| *** dina_belova has quit IRC | 00:00 | |
| mordred | jeblair: it does not print the name until the test is done - so you'll still need to see which test it's ran last before the issue | 00:00 |
|---|---|---|
| mordred | jeblair: and then look at the test list to see which one it's expecting to run next | 00:00 |
| jeblair | mordred: yeah, i had just worked that out. | 00:00 |
| * mordred just had a thought as to how to make that output better in testr too | 00:01 | |
| jeblair | mordred: that got me the answer, thanks. | 00:01 |
| jeblair | I'll add this to TESTING.rst too. | 00:01 |
| mordred | woot | 00:01 |
| mordred | sorry for the trouble - I agree, testr should give you a better experience here | 00:02 |
| jeblair | mordred: i think the thing that is surprising for me is that everything interesting seems to involve subunit, which seems like it should be under-the-hood. | 00:04 |
| mordred | jeblair: yes, well, it shold be under the hood. the places where it's surfacing are places where the ui is breaking down and we're manipulating the underlying stuff to do something | 00:05 |
| jeblair | *nod* | 00:05 |
| mordred | on a normal day-to-day basis, I never touch subunit commands or options (thus why it always takes me a second to come up with the right answer for you) | 00:05 |
| openstackgerrit | colinmcnamara proposed a change to openstack-infra/config: Add goals and docbook-properties.sh script to openstack-training https://review.openstack.org/39185 | 00:06 |
| *** mberwanger has quit IRC | 00:12 | |
| fungi | mordred: oddly, adding pbr to requirements.txt does *not* result in it being installed globally upon python setup.py install | 00:13 |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/zuul: Support multiple triggers https://review.openstack.org/39186 | 00:13 |
| mordred | fungi: ah - no - now THAT's the bug I thought we were hitting the other day | 00:13 |
| pleia2 | ok, all done for bug day \o/ https://etherpad.openstack.org/cibugreview-july2013 | 00:13 |
| mordred | that's the one where pip is finding the egg in the dir | 00:13 |
| pleia2 | fungi: clarkb ^^ | 00:14 |
| mordred | and is not doing the right thing | 00:14 |
| *** sarob_ has joined #openstack-infra | 00:14 | |
| mordred | and I belive dstufft and I have talked about ways to fix that | 00:14 |
| pleia2 | I also linked it on our InfraTeam wiki page in case we lose it ;) | 00:14 |
| pleia2 | (June one too) | 00:14 |
| dstufft | what'd I do now | 00:14 |
| fungi | mordred: well, even after clearing out the local eggs, it seems not to get any better | 00:14 |
| dstufft | oh right | 00:15 |
| dstufft | that problem | 00:15 |
| mordred | dstufft: yeah. that funness | 00:15 |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/zuul: Add a timer trigger (WIP) https://review.openstack.org/39187 | 00:15 |
| * mordred going to eat | 00:15 | |
| dstufft | don't call pip yourself ;P | 00:15 |
| fungi | i think it downloads a new one into the cwd and then sees that and uses it instead of installing globally | 00:15 |
| jeblair | that one ^ I've just started on, but the change before it may be worth a look. | 00:16 |
| *** nati_ueno has quit IRC | 00:16 | |
| dstufft | fungi: yes | 00:16 |
| *** emagana has quit IRC | 00:16 | |
| *** sarob has quit IRC | 00:17 | |
| fungi | so if we want python setup.py install to work from within the source tree, i'm afraid we probably have to back pbr out | 00:17 |
| dstufft | fungi: this doesn't affect pip except when you do what openstack does and call pip yourself inside the setup.py requires script | 00:17 |
| fungi | or else pbr needs a redesign in the way it's satisfying dependencies | 00:18 |
| *** nati_uen_ has quit IRC | 00:18 | |
| *** nati_ueno has joined #openstack-infra | 00:19 | |
| fungi | oh, though in *this* case we can at least make it work by switching to python setup.py install to pip install . | 00:20 |
| dstufft | things generally owrk better if you pretend setup.py doesn't exist and it's only an entrypoint for pip | 00:20 |
| fungi | but that doesn't satisfy jeblair's desire to preserve setup.py being callable | 00:21 |
| fungi | so we may still be at somewhat of an impass | 00:21 |
| fungi | unless setup.py's install target can call pip install . itself? or is that getting too circular | 00:22 |
| dstufft | sounds like a bad idea, setup.py install already calls pip | 00:25 |
| dstufft | because of pbr | 00:25 |
| clarkb | pleia2: thank you | 00:25 |
| dstufft | hm | 00:25 |
| dstufft | I might make a patch to one of the OS projects to show how I think it should be packaged ;P to see if it breaks anything because I think you'll end up with far less corner cases | 00:26 |
| dstufft | not sure which one | 00:26 |
| dstufft | or atleast the corner cases you get are the ones everyone is getting :V | 00:26 |
| *** gyee has quit IRC | 00:27 | |
| *** ^demon has quit IRC | 00:27 | |
| clarkb | jeblair: is there a non code reason for making triggers map 1:1 to pipelines? | 00:28 |
| *** nati_uen_ has joined #openstack-infra | 00:29 | |
| openstackgerrit | A change was merged to openstack-dev/hacking: Don't crash on tests when git is not installed https://review.openstack.org/37870 | 00:29 |
| openstackgerrit | A change was merged to openstack-dev/hacking: Make H202 work again https://review.openstack.org/39180 | 00:30 |
| openstackgerrit | A change was merged to openstack-infra/config: Serve IRC log files as MIME type text/plain. https://review.openstack.org/38050 | 00:37 |
| clarkb | I will check on the logs once that change goes into effect | 00:38 |
| *** nati_uen_ has quit IRC | 00:40 | |
| fungi | clarkb: should work. i checked that change out onto eavesdrop temporarily back when i first submitted it... worked as intended | 00:43 |
| jeblair | clarkb: depends on what you mean by code... i'm not sure where/how a pipeline with two triggers would report | 00:43 |
| clarkb | jeblair: code ~= implementation details. Your answer is what I was looking for | 00:44 |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/config: Switch Zuul/JJB to use LOG_PATH https://review.openstack.org/39133 | 00:47 |
| jeblair | clarkb, fungi: ^ found the missing file | 00:47 |
| *** nati_ueno has quit IRC | 00:47 | |
| *** pabelanger has quit IRC | 00:47 | |
| fungi | was it under the sofa cushions? | 00:49 |
| clarkb | mordred stole it | 00:49 |
| clarkb | or his goat ate it | 00:50 |
| *** nati_ueno has joined #openstack-infra | 00:51 | |
| clarkb | fungi: irc logs look good | 00:52 |
| *** nati_ueno_2 has joined #openstack-infra | 00:52 | |
| *** dkranz has joined #openstack-infra | 00:53 | |
| clarkb | woo feel like I accomplished almost nothing today | 00:54 |
| fungi | jeblair: on 39133 does openstack_functions.py get applied until zuul is restarted? will we have a race if the changes to the jjb configs get loaded before that happens and log_path doesn't look up to anything yet? | 00:55 |
| clarkb | fungi: I believe the BASE_LOG_PATH is kept in that function so that there isn't a race | 00:55 |
| clarkb | fungi: once all of the jenkins jobs are updated BASE_LOG_PATH can be removed | 00:55 |
| *** dina_belova has joined #openstack-infra | 00:55 | |
| fungi | but that change starts using log_path in the jobs right away, even if zuul isn't providing it yet (or probably i'm misreading) | 00:56 |
| *** nati_ueno_2 has quit IRC | 00:56 | |
| fungi | seems to me we'd need to stop puppet on jenkins, merge 39133, restart zuul and then let puppet loose on jenkins again | 00:57 |
| clarkb | fungi: oh I see. there is a race the opposite direction of what I was thinking | 00:57 |
| clarkb | fungi: yeah Ithink we want to make the zuul restart go in first. Unless the zuul change for making reloads reimport functions went in then we just have to make sure zuul reloads its config first | 00:58 |
| fungi | which seems probably okay of the functions extension is loaded on the fly in zuul, but if it needs to wait or a graceful restart... | 00:58 |
| fungi | ye gods i can't type | 00:58 |
| clarkb | fungi: don't worry I am afflicted with poor typing too | 00:59 |
| * clarkb AFKs and resolves to be more useful tomorrow | 01:00 | |
| *** dina_belova has quit IRC | 01:00 | |
| clarkb | though I did spend some time with zaro derping on gerrit and I think the WIP stuff may be getting close | 01:00 |
| *** dkranz has quit IRC | 01:04 | |
| *** yaguang has joined #openstack-infra | 01:11 | |
| *** hashar has quit IRC | 01:16 | |
| *** yaguang has quit IRC | 01:16 | |
| *** ^demon has joined #openstack-infra | 01:17 | |
| *** ^demon has joined #openstack-infra | 01:17 | |
| *** nati_ueno has quit IRC | 01:25 | |
| *** anteaya has quit IRC | 01:30 | |
| *** CaptTofu has joined #openstack-infra | 01:36 | |
| *** yaguang has joined #openstack-infra | 01:40 | |
| *** UtahDave has quit IRC | 01:49 | |
| *** sarob_ has quit IRC | 01:51 | |
| *** dina_belova has joined #openstack-infra | 01:56 | |
| *** ^demon has quit IRC | 01:57 | |
| *** ^demon has joined #openstack-infra | 01:58 | |
| *** ^demon has joined #openstack-infra | 01:58 | |
| *** dina_belova has quit IRC | 02:01 | |
| *** ^demon has quit IRC | 02:02 | |
| *** ^demon has joined #openstack-infra | 02:03 | |
| *** ^demon has joined #openstack-infra | 02:03 | |
| *** mirrorbox has quit IRC | 02:05 | |
| *** mirrorbox has joined #openstack-infra | 02:06 | |
| *** markmcclain has quit IRC | 02:15 | |
| mordred | clarkb: woot | 02:16 |
| *** dguitarbite has joined #openstack-infra | 02:17 | |
| *** CaptTofu has quit IRC | 02:19 | |
| *** pabelanger has joined #openstack-infra | 02:21 | |
| *** nati_ueno has joined #openstack-infra | 02:22 | |
| *** woodspa_ has quit IRC | 02:23 | |
| *** michchap has quit IRC | 02:27 | |
| mordred | dstufft, fungi: you will have a VERY hard time convincing me to back pbr out of things | 02:49 |
| mordred | fungi: I will figure out the 'get pbr installed via setup.py install' thing | 02:49 |
| fungi | mordred: i'm looking forward to it! | 02:50 |
| mordred | fungi: also - what was the issue that we had with pbr not being in the requirements.txt ? | 02:50 |
| mordred | I know I said it was a bug that it wasnt' there - but actually, it's not a runtime dep of jeepyb | 02:50 |
| fungi | mordred: entrypoints wrappers complain about not being able to find pbr when trying to run jeepyb commands | 02:51 |
| mordred | fungi: well, that's about to go away with the next version of pbr | 02:51 |
| fungi | ahh, so that dissipates with the entrypoints rework? excellent | 02:52 |
| mordred | yes | 02:52 |
| fungi | apparently also wants d2to1 and pycrypto before the entrypoints wrappers work, but doesn't install them either | 02:53 |
| fungi | and pycrypto in turn wants a cc installed so it can compile some non-python buts | 02:54 |
| fungi | bits | 02:54 |
| mordred | d2to1 is going away | 02:54 |
| mordred | and there is nothing we can do about the cc requirement to express it | 02:54 |
| fungi | yeah, the cc i get we have to preaddress | 02:54 |
| mordred | yup :) | 02:55 |
| mordred | here - let me release pbr again with d2to1 gone and entrypoint console_scripts rewritten | 02:55 |
| fungi | just wondering if pycrypto is really a dependency either in that case, or something else going away with the entrypoints rewrite | 02:55 |
| *** dina_belova has joined #openstack-infra | 02:56 | |
| *** jjmb has joined #openstack-infra | 02:56 | |
| *** dina_belova has quit IRC | 03:01 | |
| *** jjmb has quit IRC | 03:04 | |
| *** mberwanger has joined #openstack-infra | 03:08 | |
| mordred | fungi: any idea what wants pycrypto? | 03:09 |
| *** HenryG has quit IRC | 03:10 | |
| fungi | mordred: not sure. it's not in requirements so it's got to be transitive. i can remove it and see if i can divine why entrypoints errors about it missing | 03:12 |
| mordred | ok. I just released 0.5.20 | 03:15 |
| mordred | fungi: paramiko | 03:16 |
| fungi | mordred: http://paste.openstack.org/show/42556/ | 03:16 |
| fungi | mordred: ahh, possible. i was going to guess python-swiftclient | 03:17 |
| fungi | but hadn't checked it yet | 03:17 |
| mordred | fungi: so - lifeless wants run-mirror to be its own thing | 03:17 |
| mordred | fungi: I wonder if perhaps jeepyb really wants to be a bunch of different repos that all have a depend on an as-yet-un-created library to parse projects.yaml | 03:18 |
| mordred | because really, other than that file, these programs have no shared purpose in life | 03:18 |
| *** sarob has joined #openstack-infra | 03:18 | |
| fungi | i've pondered that from time to time and come to a similar conclusion | 03:19 |
| fungi | but this goes back to convenience. we put a lot of these things in the config repo originally because we didn't want to make a ton of tiny little separately-packaged projects | 03:19 |
| fungi | then forklifted them into jeepyb for similar convenience reasons, but to still get them out of the config repo | 03:20 |
| fungi | but also that was back before we had quite as much automation around managing and releasing projects... so maybe that's less inconvenient at this point? | 03:21 |
| openstackgerrit | A change was merged to openstack/requirements: Clean up pass-through logic in update.py https://review.openstack.org/36037 | 03:23 |
| lifeless | mordred: well, I don't want devs to have to install libmysql-dev etc to get going | 03:24 |
| *** melwitt has quit IRC | 03:25 | |
| fungi | mordred: so anyway, back to the paramiko/pycrypto argument. apparently setup.py install in jeepyb now fails to pull in that transitive dependency as well, and i guess the underlying problem for that isn't going away with the entrypoints rewrite? | 03:25 |
| mordred | wait - hangon | 03:26 |
| mordred | you're saying that jeepyb is failing to install pycrypto even if you do have the c stuff around? | 03:26 |
| fungi | mordred: yes | 03:26 |
| fungi | unless i pip install . instead of python setup.py install | 03:27 |
| mordred | fungi: I do not get that behavior | 03:27 |
| mordred | fungi: I just did "virtualenv foo; foo/bin/python setup.py install" and it's installing pycrypto | 03:27 |
| fungi | too fun. this is as root in the global system context on centos 6.4, fwiw | 03:28 |
| fungi | so also python 2.6 | 03:29 |
| mordred | fungi: which machine? | 03:29 |
| mordred | git.o.o ? | 03:29 |
| fungi | git.openstack.org, yes | 03:29 |
| fungi | i've since worked around it manually by pip installing pbr, d2to1 and pycrypto from pypi | 03:30 |
| mordred | ok | 03:30 |
| fungi | but can repeat it by uninstalling those | 03:30 |
| fungi | actually i've not reinstalled pycrypto since uninstalling it a minute ago to get a paste of the entrypoints error | 03:32 |
| mordred | fungi: I just did, so that I could watch what happens | 03:35 |
| fungi | i'm not doing anything with it at the moment, so troubleshoot what you like | 03:36 |
| *** mberwang_ has joined #openstack-infra | 03:45 | |
| *** afazekas__ has quit IRC | 03:48 | |
| *** mberwanger has quit IRC | 03:48 | |
| Ryan_Lane | hey folks, since I've spent all day remediating BEAST on a shit-ton of misc services, I thought I'd talk to you guys about it | 03:54 |
| clarkb | BEAST? I clealy must lwarn more | 03:54 |
| Ryan_Lane | https://www.ssllabs.com/ssltest/analyze.html?d=review.openstack.org | 03:55 |
| clarkb | oh that | 03:55 |
| Ryan_Lane | you need to limit your ciphers and need to set a server preference | 03:55 |
| Ryan_Lane | https://gerrit.wikimedia.org/r/#/c/76679/ | 03:55 |
| Ryan_Lane | I just did it to 17 services or so | 03:55 |
| clarkb | that looks like something we should add to our https vhosts | 03:57 |
| Ryan_Lane | yep | 03:57 |
| *** dina_belova has joined #openstack-infra | 03:57 | |
| Ryan_Lane | there's still an open SSL vulnerability with RC4, but it's apparently less practical than BEAST | 03:58 |
| Ryan_Lane | so SSL is actually vulnerable till TLS 1.2 is properly supported everywhere | 03:58 |
| *** dina_belova has quit IRC | 04:02 | |
| *** afazekas has joined #openstack-infra | 04:03 | |
| clarkb | etherpad, wiki, review, and jenkins are the vhosts that need updating I think | 04:03 |
| * Ryan_Lane nods | 04:03 | |
| fungi | well, for the most part beast is solved in modern web browsers, and the technique is infeasible in other ssl-using protocols besides http | 04:03 |
| fungi | but shoring up the servers to cater to browsers which are vulnerable isn't a terrible idea | 04:04 |
| Ryan_Lane | the config I have doesn't include forward secrecy ciphers, btw | 04:04 |
| Ryan_Lane | if you care about those, you may want to add them back in | 04:05 |
| Ryan_Lane | they are relatively expensive | 04:05 |
| *** yamahata has quit IRC | 04:05 | |
| Ryan_Lane | (those are ciphers with DHE and ECDHE) | 04:05 |
| Ryan_Lane | fungi: indeed. it's still a problem in places | 04:06 |
| clarkb | do browsers do pfs regularly? | 04:06 |
| Ryan_Lane | if they support it they'll use it | 04:06 |
| Ryan_Lane | if they and the server, that is | 04:06 |
| fungi | and good that you're putting a gcm mode before rc4, given the recently discussed weaknesses | 04:11 |
| mordred | fungi, clarkb, Ryan_Lane: you know - it would be great if, in all of our puppet, we were better about having some of the repeated apache config chunks, well, modularized | 04:12 |
| mordred | cause right now it's just cargo-culted all the hell over the place | 04:12 |
| clarkb | ya :( | 04:13 |
| fungi | mordred: well, and in places we use the default vhost provided by the module too | 04:13 |
| mordred | fungi: yah. | 04:13 |
| mordred | I mean, it's working fine, but it seems like an area in which a thing like puppet sohuld really be able to lend us some consistency and ease of use | 04:14 |
| fungi | which is not very configurable without replacing, at least in our version of the module | 04:14 |
| mordred | yah | 04:14 |
| fungi | but we're also on a fairly old version of it, if memory serves | 04:14 |
| fungi | yay! pbr and oslo.config py33 tests pass on all four of the production precisepy3k slaves | 04:15 |
| clarkb | yes our version is ancient | 04:15 |
| clarkb | fungi nice. do we have working py3k slaves now? | 04:16 |
| clarkb | jd__ can start using them? | 04:16 |
| *** yamahata has joined #openstack-infra | 04:17 | |
| mordred | sdague: when you wake up ... | 04:18 |
| fungi | clarkb: yep, just reversed my -2 on https://review.openstack.org/37426 to a +2 now | 04:18 |
| openstackgerrit | Monty Taylor proposed a change to openstack/requirements: Update update.py to also write a stock setup.py https://review.openstack.org/39206 | 04:19 |
| openstackgerrit | Monty Taylor proposed a change to openstack/requirements: Update pep8 exclusion https://review.openstack.org/39207 | 04:19 |
| mordred | sdague: ^^ look at those | 04:19 |
| mordred | sdague: also, we're going to need to land https://review.openstack.org/#/c/30522/ before the update-everything-automatically will work | 04:20 |
| *** krtaylor has joined #openstack-infra | 04:20 | |
| clarkb | fungi: I think jog0 and hacking were also interested in py3k jobs | 04:21 |
| fungi | clarkb: the more the merrier! | 04:21 |
| fungi | but i suspect four will be enough initially | 04:22 |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Test to ensure the include_package_data works https://review.openstack.org/38326 | 04:22 |
| clarkb | ya and starting small is probably a good thing | 04:22 |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Cleaned up an evil carryover from the past https://review.openstack.org/39208 | 04:26 |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Test to ensure the include_package_data works https://review.openstack.org/38326 | 04:26 |
| mordred | clarkb: ok. I fixed your code review questions above ^^ | 04:27 |
| *** dguitarbite has quit IRC | 04:27 | |
| *** dguitarbite has joined #openstack-infra | 04:29 | |
| clarkb | mordred: reviewed | 04:32 |
| mordred | yay! | 04:32 |
| mordred | clarkb: sometimes I don't like you | 04:32 |
| clarkb | sorry | 04:33 |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Cleaned up an evil carryover from the past https://review.openstack.org/39208 | 04:33 |
| mordred | :) | 04:33 |
| *** yamahata has quit IRC | 04:34 | |
| *** dguitarbite has quit IRC | 04:36 | |
| *** yamahata has joined #openstack-infra | 04:36 | |
| clarkb | mordred: http://tox.im/ | 04:36 |
| clarkb | the name overloading is not very nice | 04:37 |
| mordred | wow | 04:37 |
| mordred | clarkb: let me know if I can ever install it | 04:40 |
| *** vipul is now known as vipul-away | 04:41 | |
| *** SergeyLukjanov has joined #openstack-infra | 04:42 | |
| mordred | dstufft: I sometimes feel like you just sit around waiting for us to tell you one more thing pbr does that makes you want to kill someone ;) | 04:42 |
| dstufft | mordred: working with MANIFEST files would be a reasonable thing for pbr to do :) | 04:43 |
| mordred | heh | 04:43 |
| dstufft | or getting setuptools to do it using git or whatever | 04:43 |
| mordred | yeah. basically, it does the same as setuptools-git does | 04:43 |
| dstufft | I jsut hadn't remembered seeing it do that, and a cursory glance at the code didn't make it stand out that it did it | 04:44 |
| mordred | except the hook point in setuptools for actually doing that | 04:44 |
| mordred | is in a crazy place | 04:44 |
| mordred | so it runs it EVEN IF YOU ARE INSTALLING FROM A TARBALL | 04:44 |
| mordred | which is almost never what you want | 04:44 |
| dstufft | lolsetuptools | 04:44 |
| mordred | dstufft: oh - sorry, that patch hasn't landed yet | 04:44 |
| dstufft | mordred: ah | 04:44 |
| mordred | dstufft: https://review.openstack.org/#/c/30522/ | 04:44 |
| dstufft | so I'm not crazy (for that reason at least ;) ) | 04:44 |
| mordred | yeah - I mean, let's not go too far :) | 04:45 |
| *** nibalizer has quit IRC | 04:46 | |
| *** vipul-away is now known as vipul | 04:47 | |
| *** nibalizer has joined #openstack-infra | 04:47 | |
| dstufft | I generally just maintain my MANIFEST.in by hand, but i'm a crazy person who actually understands how it's used and it isn't just "something I try when install stops working" | 04:47 |
| mordred | yah | 04:48 |
| mordred | well, we started that way | 04:48 |
| mordred | and made more than one release where important files were missing from the tarball | 04:48 |
| mordred | *doh* | 04:48 |
| dstufft | I also don't have like 30 different projects with tons of different people committing on them either so | 04:48 |
| * mordred has missed the make distcheck target from autotools on this one | 04:48 | |
| mordred | :) | 04:48 |
| * mordred has almost been tempeted to write a python equiv to make distcheck | 04:49 | |
| mordred | but it seems like the path to madness perhaps | 04:49 |
| *** melwitt has joined #openstack-infra | 04:49 | |
| dstufft | mordred: I've barely used C I forget what that does :[ Assumingly checks a packaged tool? | 04:49 |
| dstufft | s/tool/tarball-whatever/ | 04:49 |
| mordred | it make a dist tarball, then unpacks that into a directory, then checks that the source can build in a vpath build adjacent to that directory, then ensures that the build doesn't try to write files into the source tree inappropriately, then checks that make clean cleans up | 04:50 |
| dstufft | ah | 04:51 |
| mordred | (build doesn't try part is that it sets the source dir readonly, but since it's doing a vpath build, which means alternate build dir, there is a place for them to go ... so it checks that all built files go into $builddir | 04:51 |
| dstufft | I think people should run their tests against their releases prior to officially releases them | 04:51 |
| mordred | it's super civilized | 04:51 |
| dstufft | releasing | 04:51 |
| mordred | yes. that's actually on my list | 04:52 |
| mordred | I want to install everything like we do for devstack - and then run all the unittest suites against those installations without the source dir around | 04:52 |
| dstufft | awesome :3 | 04:54 |
| clarkb | if we could just get people to test things that would be a big win | 04:56 |
| clarkb | but in this particular cases testing the sdists before pushing to pypi (or wherever) would definitely eb a good thing | 04:56 |
| * mordred hands clarkb a good luck cookie on people testing | 04:56 | |
| mordred | I do not think we should test them before pushing to pypi | 04:57 |
| mordred | I think we shoudl test them on every commit | 04:57 |
| mordred | but yes | 04:57 |
| mordred | we DO actually test them to a degree now... | 04:57 |
| *** dina_belova has joined #openstack-infra | 04:57 | |
| clarkb | mordred: to a very high degree if you are using version control as your source of versions | 04:57 |
| mordred | the pbr integration test does test that all projects can at least install into a venv from their tarball | 04:57 |
| clarkb | which I think is a great idea because that is what version control exists for | 04:57 |
| mordred | clarkb: ++ | 04:58 |
| clarkb | unfortunately the convert a version to an installable thing is the tricky bit | 04:58 |
| mordred | bah. piece of cake :) | 04:59 |
| dstufft | mordred: I've been debating crazy hacks to wrest control of setup_requires from setuptools and give it to pip | 04:59 |
| mordred | dstufft: I would consider you an absoulte hero | 04:59 |
| mordred | dstufft: easier might be - get setuptools to just write the contents into the egg-info dir | 05:00 |
| mordred | so that if it's there, and pip sees it, pip will install those _before_ running python setup.py egg_info | 05:00 |
| mordred | really just helps with released tarballs I guess... | 05:00 |
| mordred | but it would be a start? | 05:01 |
| mordred | or - you could just wrest control | 05:01 |
| dstufft | yea. I have some ideas. None particularly clean but then again it's setuptools | 05:01 |
| mordred | dstufft: I support all of them | 05:01 |
| *** Ryan_Lane1 has joined #openstack-infra | 05:01 | |
| mordred | dstufft: just earlier today | 05:01 |
| mordred | I actually considered vendor-including setuptools in pbr | 05:02 |
| mordred | so that I could make it do evil things | 05:02 |
| mordred | but it turns out it wouldn't have helped | 05:02 |
| dstufft | heh | 05:02 |
| *** dina_belova has quit IRC | 05:02 | |
| dstufft | I've been debating attempting to vendor setuptools in pip | 05:02 |
| dstufft | turns out that's actually pretty hard | 05:02 |
| *** Ryan_Lane has quit IRC | 05:02 | |
| mordred | well... | 05:02 |
| *** Ryan_Lane1 is now known as Ryan_Lane | 05:02 | |
| *** Ryan_Lane has joined #openstack-infra | 05:02 | |
| dstufft | setuptools main trick requires a global .pth back | 05:03 |
| mordred | what about merging the two trees, having setuptools always install pip | 05:03 |
| dstufft | hack | 05:03 |
| mordred | and then deleting easy_install from the source | 05:03 |
| mordred | (I know, I know - windows apparently) | 05:03 |
| *** mberwang_ has quit IRC | 05:03 | |
| dstufft | windows actually isn't that important anymore with wheels | 05:04 |
| dstufft | mostly just getting the community onto wheels is the problem there | 05:05 |
| dstufft | mostly we want to demphasize setuptools | 05:05 |
| dstufft | With wheels (other than a small issue with the auto generated script wrappers) you can theortically install without setuptools at all | 05:05 |
| dstufft | we need a sdist format that offers the same ability, and move setuptools to a build only tool | 05:06 |
| dstufft | e.g. you might use setuptools to turn a sdist into a wheel, but openstack could, for instance, turn pbr into a full blown build tool that makes sdists and wheels | 05:06 |
| *** michchap has joined #openstack-infra | 05:08 | |
| Ryan_Lane | mordred: dude, you think openstack ci's puppet repo is bad about apache? | 05:10 |
| Ryan_Lane | wikimedia's makes me want to strangle myself | 05:10 |
| mordred | dstufft: I am very looking forward to that | 05:10 |
| mordred | Ryan_Lane: I do not! | 05:10 |
| Ryan_Lane | tackle your apache problems when they are small or have like 4 different ways to handle apache with like 50 services | 05:11 |
| mordred | Ryan_Lane: we actually went through and got buy in from prior contribs | 05:11 |
| mordred | oh | 05:11 |
| mordred | I thought you were talking about license | 05:11 |
| Ryan_Lane | oh | 05:11 |
| Ryan_Lane | heh | 05:11 |
| Ryan_Lane | nah | 05:11 |
| mordred | Ryan_Lane: yes - I'm sure yours are worse! | 05:11 |
| Ryan_Lane | we have no consistent way of doing it and nothing is reusable | 05:11 |
| mordred | but right? shouldn't the apache shit just be, you know, templated? | 05:11 |
| Ryan_Lane | apache is actually one of the harder ones | 05:11 |
| Ryan_Lane | because it's used for so much random crap, all of which has wildly different config | 05:12 |
| Ryan_Lane | and puppetlabs apache module is a giant piece of crap | 05:12 |
| clarkb | this is where I suggest nginx to bother mordred | 05:12 |
| * mordred doesn't care at all, actually, nginx v. apache | 05:12 | |
| clarkb | Ryan_Lane: it has gotten a lot better recently | 05:12 |
| mordred | just as long as we don't use both | 05:12 |
| Ryan_Lane | we use both :) | 05:12 |
| mordred | heh | 05:12 |
| clarkb | Ryan_Lane: hunner has put a lot of work into making the apache module suck less | 05:13 |
| * mordred knows there will be variation - but you know - the ssl stuff? all should be the same | 05:13 | |
| Ryan_Lane | well, I hope so, because it did basically nothing before | 05:13 |
| Ryan_Lane | heh | 05:13 |
| mordred | the vhost stuff? all should be the same | 05:13 |
| Ryan_Lane | funny enough, we use nginx for ssl termination and that's about it | 05:13 |
| mordred | what the vhost SERVES? sure. wildly different | 05:13 |
| Ryan_Lane | and we're considering switching back to apache with 2.4 | 05:13 |
| mordred | krow was suggesting pound for ssl termination | 05:13 |
| Ryan_Lane | I keep hearing openstack folks recommending pound and I just can't wrap my head around it | 05:14 |
| Ryan_Lane | it's so, so slow | 05:14 |
| dstufft | I think a stripped down apache is still the fastest option for SSL termination | 05:14 |
| Ryan_Lane | 2.4 yes, 2.2 no ;) | 05:14 |
| * mordred doesn't actually have an opinion on pound - just parroting | 05:14 | |
| Ryan_Lane | 2.4 has a distributed ssl cache implementation \o/ | 05:14 |
| dstufft | turns out event driven archs kinda suck for TLS termination | 05:15 |
| *** nati_ueno has quit IRC | 05:16 | |
| *** nati_ueno has joined #openstack-infra | 05:18 | |
| *** vogxn has joined #openstack-infra | 05:19 | |
| *** ^demon is now known as ^demon|zzz | 05:20 | |
| clarkb | clearly we should use cherokee for everythin | 05:20 |
| *** emagana has joined #openstack-infra | 05:23 | |
| *** sarob has quit IRC | 05:23 | |
| *** sarob has joined #openstack-infra | 05:24 | |
| mordred | clarkb: or afs | 05:24 |
| *** sarob_ has joined #openstack-infra | 05:25 | |
| *** afazekas is now known as afazekas__ | 05:25 | |
| clarkb | mordred: if we had a giant afs filesystem I could point elasticsearch at that and stop worrying about running out of disk space right? | 05:25 |
| mordred | clarkb: yes | 05:25 |
| *** sarob has quit IRC | 05:26 | |
| clarkb | we should do that then | 05:26 |
| mordred | clarkb: you'd have to start worrying about adding volume servers to the afs cell | 05:26 |
| mordred | clarkb: but you could do that at the volserver level | 05:26 |
| clarkb | just make it semi aware and have it request its own cinder volumes as necessary | 05:27 |
| clarkb | then hope that network trouble doesn't break afs on top of cinder | 05:27 |
| clarkb | we run a lot of tests | 05:29 |
| clarkb | 5553 over the last 24 hours actually completely and were not prematurely killed by zuul | 05:30 |
| *** afazekas__ has quit IRC | 05:30 | |
| *** nati_ueno has quit IRC | 05:31 | |
| *** nicedice has quit IRC | 05:32 | |
| *** sarob_ has quit IRC | 05:36 | |
| *** sarob has joined #openstack-infra | 05:36 | |
| mordred | clarkb: https://review.openstack.org/Documentation/config-gerrit.html#_a_id_changemerge_a_section_changemerge | 05:38 |
| *** dguitarbite has joined #openstack-infra | 05:38 | |
| mordred | clarkb: that would be load for no purpose | 05:38 |
| mordred | but | 05:38 |
| mordred | if it could be changed to not only enable or disable the submit button - but display, oh, I dunno, a message | 05:39 |
| mordred | that would be neat | 05:39 |
| *** devananda has quit IRC | 05:39 | |
| mordred | clarkb: woah!: https://gerrit-review.googlesource.com/37460 | 05:40 |
| *** sarob has quit IRC | 05:40 | |
| clarkb | zuul tests that for us sort of | 05:41 |
| clarkb | also nice on multi master | 05:41 |
| mordred | clarkb: yeah - but not in a way that a person can see if they're just checking on their patch | 05:42 |
| *** emagana has quit IRC | 05:42 | |
| *** jjmb has joined #openstack-infra | 05:46 | |
| *** jjmb has quit IRC | 05:48 | |
| *** afazekas has joined #openstack-infra | 05:49 | |
| *** dguitarbite has left #openstack-infra | 05:53 | |
| *** emagana has joined #openstack-infra | 05:58 | |
| *** dina_belova has joined #openstack-infra | 05:58 | |
| openstackgerrit | Jay Lau proposed a change to openstack-dev/hacking: Update I18N code style https://review.openstack.org/39217 | 05:59 |
| *** pcrews has joined #openstack-infra | 05:59 | |
| *** dina_belova has quit IRC | 06:03 | |
| *** michchap has quit IRC | 06:04 | |
| *** michchap has joined #openstack-infra | 06:05 | |
| *** emagana has quit IRC | 06:08 | |
| *** SergeyLukjanov has quit IRC | 06:10 | |
| *** michchap has quit IRC | 06:28 | |
| *** emagana has joined #openstack-infra | 06:41 | |
| *** sarob has joined #openstack-infra | 06:47 | |
| *** emagana has quit IRC | 06:47 | |
| *** Ryan_Lane has quit IRC | 06:50 | |
| *** nati_ueno has joined #openstack-infra | 06:51 | |
| *** sarob has quit IRC | 06:52 | |
| *** Ryan_Lane has joined #openstack-infra | 06:52 | |
| *** melwitt has quit IRC | 06:59 | |
| *** sarob has joined #openstack-infra | 07:09 | |
| *** SergeyLukjanov has joined #openstack-infra | 07:10 | |
| *** nati_ueno has quit IRC | 07:10 | |
| openstackgerrit | Julien Danjou proposed a change to openstack-infra/config: Add pycadf project https://review.openstack.org/39225 | 07:21 |
| *** vogxn has quit IRC | 07:31 | |
| *** michchap has joined #openstack-infra | 07:33 | |
| *** sarob has quit IRC | 07:41 | |
| *** sarob has joined #openstack-infra | 07:42 | |
| *** jpich has joined #openstack-infra | 07:42 | |
| *** nayward has joined #openstack-infra | 07:43 | |
| *** nayward has quit IRC | 07:44 | |
| *** nayward has joined #openstack-infra | 07:44 | |
| *** sarob has quit IRC | 07:46 | |
| *** vogxn has joined #openstack-infra | 07:48 | |
| *** emagana has joined #openstack-infra | 07:58 | |
| *** dina_belova has joined #openstack-infra | 07:59 | |
| *** psedlak has joined #openstack-infra | 08:01 | |
| *** SergeyLukjanov has quit IRC | 08:02 | |
| *** emagana has quit IRC | 08:03 | |
| *** dina_belova has quit IRC | 08:03 | |
| *** jjmb has joined #openstack-infra | 08:05 | |
| *** SergeyLukjanov has joined #openstack-infra | 08:05 | |
| *** dina_belova has joined #openstack-infra | 08:14 | |
| *** odyssey4me has joined #openstack-infra | 08:14 | |
| *** hashar has joined #openstack-infra | 08:20 | |
| openstackgerrit | Julien Danjou proposed a change to openstack-dev/pbr: Use the same logic everywhere for requirements files https://review.openstack.org/38540 | 08:21 |
| *** hashar has quit IRC | 08:21 | |
| *** hashar has joined #openstack-infra | 08:21 | |
| *** Ryan_Lane has quit IRC | 08:30 | |
| *** jjmb has quit IRC | 08:33 | |
| *** derekh has joined #openstack-infra | 08:38 | |
| *** jjmb has joined #openstack-infra | 08:40 | |
| *** odyssey4me has quit IRC | 08:40 | |
| *** jjmb has quit IRC | 08:40 | |
| *** odyssey4me has joined #openstack-infra | 08:48 | |
| *** nayward has quit IRC | 08:49 | |
| *** sarob has joined #openstack-infra | 08:52 | |
| *** koolhead17 has quit IRC | 08:57 | |
| *** sarob has quit IRC | 08:57 | |
| *** nayward has joined #openstack-infra | 09:05 | |
| *** odyssey4me has quit IRC | 09:05 | |
| *** yaguang has quit IRC | 09:06 | |
| *** dina_belova has quit IRC | 09:08 | |
| *** nayward has quit IRC | 09:09 | |
| *** yaguang has joined #openstack-infra | 09:15 | |
| *** dina_belova has joined #openstack-infra | 09:39 | |
| *** dina_belova has quit IRC | 09:50 | |
| *** michchap has quit IRC | 10:04 | |
| *** michchap has joined #openstack-infra | 10:04 | |
| *** nayward has joined #openstack-infra | 10:13 | |
| *** dina_belova has joined #openstack-infra | 10:15 | |
| *** nayward has quit IRC | 10:23 | |
| *** nayward has joined #openstack-infra | 10:30 | |
| *** jjmb has joined #openstack-infra | 10:34 | |
| *** nayward has quit IRC | 10:35 | |
| *** vogxn has quit IRC | 10:45 | |
| *** dina_belova has quit IRC | 10:49 | |
| *** yaguang has quit IRC | 10:56 | |
| *** nayward has joined #openstack-infra | 10:57 | |
| BobBall | sdague: If you're around, got a min to talk about my change with devstack and removing python-lxml? | 10:57 |
| *** odyssey4me has joined #openstack-infra | 11:03 | |
| *** dina_belova has joined #openstack-infra | 11:15 | |
| openstackgerrit | David Caro proposed a change to openstack-infra/jenkins-job-builder: Add support for the Summary Report publisher https://review.openstack.org/39264 | 11:16 |
| openstackgerrit | David Caro proposed a change to openstack-infra/jenkins-job-builder: Add support for the Summary Report publisher https://review.openstack.org/39264 | 11:18 |
| *** zul has quit IRC | 11:18 | |
| *** zul has joined #openstack-infra | 11:18 | |
| openstackgerrit | David Caro proposed a change to openstack-infra/jenkins-job-builder: Add support for the Summary Report publisher https://review.openstack.org/39264 | 11:31 |
| *** zul has quit IRC | 11:53 | |
| *** zul has joined #openstack-infra | 11:55 | |
| *** zul has quit IRC | 12:01 | |
| *** rfolco has joined #openstack-infra | 12:01 | |
| *** hashar has quit IRC | 12:03 | |
| BobBall | Hmmmm | 12:07 |
| BobBall | can I recheck a change after it has been approved? | 12:07 |
| BobBall | I don't want to reverify it because there _might_ be an issue | 12:07 |
| BobBall | I want to recheck a few times first | 12:07 |
| BobBall | I tried it - but it didn't seem to add it to the queue | 12:07 |
| *** zul has joined #openstack-infra | 12:10 | |
| *** cppcabrera has joined #openstack-infra | 12:16 | |
| *** cppcabrera has left #openstack-infra | 12:16 | |
| *** zul has quit IRC | 12:24 | |
| *** zul has joined #openstack-infra | 12:24 | |
| *** sandywalsh has quit IRC | 12:26 | |
| fungi | BobBall: reverify | 12:27 |
| BobBall | but I'd rather it didn't merge even if it succeeds :) | 12:27 |
| BobBall | It's a strange one | 12:27 |
| fungi | oh, nevermind | 12:28 |
| BobBall | a timeout bug that only I have seen on this change | 12:28 |
| fungi | just read what you actually wrote | 12:28 |
| BobBall | I last saw it on changeset #1 once - a recheck made it disappear and I didn't see it on changeset's 2 or 3 | 12:28 |
| BobBall | but the gate saw it | 12:28 |
| BobBall | and nobody has seen it in the mean time | 12:28 |
| BobBall | so I'm very confused | 12:28 |
| fungi | you could upload a similar change with a separate change id | 12:28 |
| BobBall | I suppose - but I wonder if it's something more subtle than that... My change itself doesn't touch anything that should be around the DB migration code (I suppose that a link to the change would be good... https://review.openstack.org/#/c/36668/ ) | 12:29 |
| BobBall | I'll upload an identical change now | 12:30 |
| *** dina_belova has quit IRC | 12:34 | |
| jpich | BobBall: I think "recheck" comments only are taken into account if they're the only thing in the comment. Since you also wrote an explanation it probably wasn't picked up | 12:35 |
| BobBall | d'oh... | 12:35 |
| BobBall | I'll recheck that too | 12:35 |
| *** sandywalsh has joined #openstack-infra | 12:40 | |
| BobBall | jpich: you're right - the recheck is running now | 12:41 |
| *** dkranz has joined #openstack-infra | 12:42 | |
| jpich | BobBall: Cool! | 12:43 |
| *** w_ is now known as olaph | 12:43 | |
| jpich | Neat little progress bars on zuul, hadn't seen that before | 12:43 |
| *** dina_belova has joined #openstack-infra | 12:49 | |
| mordred | jpich: they're new! | 12:58 |
| jpich | mordred: they're shiny! Great addition | 12:59 |
| *** anteaya has joined #openstack-infra | 13:01 | |
| *** mriedem has joined #openstack-infra | 13:11 | |
| *** jjmb has quit IRC | 13:11 | |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Replace setuptools_git with a smarter approach https://review.openstack.org/30522 | 13:11 |
| *** odyssey4me has quit IRC | 13:11 | |
| *** changbl has quit IRC | 13:16 | |
| *** prad_ has joined #openstack-infra | 13:28 | |
| *** hashar has joined #openstack-infra | 13:29 | |
| sdague | BobBall: around now | 13:29 |
| *** mestery has quit IRC | 13:32 | |
| sdague | mordred: commented on your reviews | 13:33 |
| sdague | also, you got a fix for the oslo thing? | 13:33 |
| *** mestery has joined #openstack-infra | 13:33 | |
| sdague | because if so, we could get that other devstack thing sorted | 13:33 |
| *** vijendar has joined #openstack-infra | 13:34 | |
| *** krtaylor has quit IRC | 13:36 | |
| openstackgerrit | Monty Taylor proposed a change to openstack/requirements: Update pep8 exclusion https://review.openstack.org/39207 | 13:36 |
| openstackgerrit | Monty Taylor proposed a change to openstack/requirements: Update update.py to also write a stock setup.py https://review.openstack.org/39206 | 13:36 |
| mordred | sdague: thanks! updated | 13:36 |
| *** emagana has joined #openstack-infra | 13:36 | |
| *** dkliban has quit IRC | 13:37 | |
| *** CaptTofu has joined #openstack-infra | 13:37 | |
| mordred | sdague: you know what I love? when a unittest works on my laptop, but not on a build slave | 13:38 |
| mordred | sdague: and when I log in to the build slave and try it by hand there, and it in fact, does not work when I run it | 13:38 |
| sdague | there were a few of those this morning | 13:38 |
| mordred | sdague: but when I run thecode by hand, I can verify that it is working | 13:38 |
| mordred | I REALLY love it | 13:38 |
| sdague | oh, fun, environment var? | 13:38 |
| *** _TheDodd_ has joined #openstack-infra | 13:39 | |
| mordred | god I hope not | 13:39 |
| sdague | that's always my assumption in those cases | 13:39 |
| sdague | but it's not always a good assumption | 13:39 |
| BobBall | sdague: Okay - so - using yum to remove python-lxml will cause all packages that depend on python-lxml to be removed. That's quite an anti-social thing to do in any case, but in the specific case I care about this causes the removal of xenserver-core because xenserver-core depends on python-lxml. The reason for removing it is because we're about to install it again via pip and we didn't want the system to have mismatching versions. | 13:40 |
| *** nayward has quit IRC | 13:41 | |
| BobBall | sdague: Perhaps we shouldn't be removing it at all, letting the packaging system think it's still installed, and over-write the files using pip install? That's what it was trying to avoid, but we need both packaging systems to know that this is installed | 13:42 |
| mordred | why do we want non-package-system lxml? | 13:42 |
| BobBall | pip handles it - presumably so we can specify requires for certain versions | 13:43 |
| mordred | this is in devstack, right? | 13:43 |
| BobBall | yes | 13:43 |
| BobBall | Based on my change at https://review.openstack.org/#/c/39100/ | 13:43 |
| BobBall | I'd love to get the github link showing the lines of interest, but it's down for me | 13:44 |
| BobBall | you can see them on the change ;) | 13:44 |
| BobBall | oh - and finally github responded... | 13:44 |
| BobBall | https://github.com/openstack-dev/devstack/blob/master/stack.sh#L601 | 13:45 |
| mordred | hrm. I'm torn on this | 13:45 |
| BobBall | I know what you mean | 13:46 |
| BobBall | it's horrible | 13:46 |
| BobBall | but comes from us wanting to use pip in conjunction with the distro's own packaging system | 13:47 |
| *** hashar_ has joined #openstack-infra | 13:47 | |
| mordred | yeah. which itself is a bit of an abominatoin | 13:48 |
| *** nayward has joined #openstack-infra | 13:48 | |
| sdague | so in all the other cases we just blow things over the distro packages | 13:48 |
| BobBall | 'fraid so, yes. If they are installed | 13:48 |
| mordred | sdague, jeblair: can I make a TERRIBLE TERRIBLE TERRIBLE suggestion? (and please feel free to kill me with sticks) | 13:49 |
| sdague | honestly, I'm more concerned that by removing packages in the required chain someone does a yum update on their box, it fills back in all the holes, then things get lots of fun | 13:49 |
| *** pabelanger_ has joined #openstack-infra | 13:49 | |
| mordred | what if we went back to producing and installing distro packages ... EXCEPT ... not rich distro packages, just ones that do the equiv to 'python setup.py install' | 13:50 |
| BobBall | sdague: Good point - so my proposed fix isn't the right one | 13:50 |
| * mordred hides | 13:50 | |
| sdague | mordred: you want to do that on rhel6? | 13:50 |
| *** hashar has quit IRC | 13:50 | |
| *** hashar_ is now known as hashar | 13:50 | |
| mordred | why not? I'm not talking about _good_ distro packages | 13:50 |
| mordred | that do _anything_ | 13:50 |
| sdague | because I didn't think you were that much of a masochist | 13:51 |
| mordred | I'm talking about things that are largely equiv to "python setup.py bdist_rpm" | 13:51 |
| sdague | oh, metadata packages | 13:51 |
| sdague | hmmm.... | 13:51 |
| mordred | yeah. I mean, they should install the python library | 13:51 |
| mordred | but not, you know, try to set config things or meet policy or any of those things | 13:51 |
| sdague | well I've got a meeting to run off to, so I'll let that rattle around | 13:51 |
| mordred | yeah - it's not that I wnat to install a system with these - but we're fighting pip dependency math so much | 13:52 |
| sdague | yeh, the bigger issue is because devstack works on a range of distros | 13:52 |
| mordred | the time spent in occasionally updating a spec and a control file to list a dep might be less work | 13:52 |
| *** michchap has quit IRC | 13:52 | |
| sdague | it installs different pips based on versions, or if you have the cloud archive there | 13:52 |
| mordred | we could have devstack build the rpms/debs as first step | 13:52 |
| sdague | so the overwrite just seems safest | 13:52 |
| sdague | anyway, meeting | 13:53 |
| mordred | k | 13:53 |
| sdague | and pondering.... | 13:53 |
| BobBall | I'm really keen that we don't remove packages that the system thinks is installed... | 13:54 |
| *** markmcclain has joined #openstack-infra | 13:55 | |
| *** markmcclain has quit IRC | 13:55 | |
| *** pabelanger_ has quit IRC | 13:58 | |
| *** pabelanger_ has joined #openstack-infra | 13:58 | |
| *** pabelanger has quit IRC | 13:58 | |
| *** pabelanger_ is now known as pabelanger | 13:58 | |
| *** pabelanger_ has joined #openstack-infra | 13:58 | |
| *** cp16net is now known as cp16net|away | 14:08 | |
| *** mrodden has joined #openstack-infra | 14:09 | |
| *** dina_belova has quit IRC | 14:10 | |
| *** krtaylor has joined #openstack-infra | 14:18 | |
| *** devananda has joined #openstack-infra | 14:18 | |
| *** dkranz has quit IRC | 14:19 | |
| *** markmcclain has joined #openstack-infra | 14:20 | |
| *** cp16net|away is now known as cp16net | 14:24 | |
| *** burt has joined #openstack-infra | 14:31 | |
| *** dkliban has joined #openstack-infra | 14:31 | |
| *** datsun180b has joined #openstack-infra | 14:33 | |
| *** dkranz has joined #openstack-infra | 14:34 | |
| *** changbl has joined #openstack-infra | 14:38 | |
| openstackgerrit | David Caro proposed a change to openstack-infra/jenkins-job-builder: Added some more scm options https://review.openstack.org/39298 | 14:38 |
| *** derekh has quit IRC | 14:39 | |
| ekarlso- | jaypipes: here ? | 14:43 |
| *** dhellmann_ is now known as dhellmann | 14:44 | |
| openstackgerrit | Roman Podolyaka proposed a change to openstack-infra/config: Modify running of sqlalchemy-migrate tests https://review.openstack.org/39304 | 14:50 |
| Shrews | mordred, fungi: not sure if you guys saw this, but if you're using java-gearman-service as part of jenkins anywhere, the project has been abandoned. http://code.google.com/p/java-gearman-service | 14:51 |
| fungi | jeblair: zaro: ^ ? | 14:52 |
| *** boris-42 has joined #openstack-infra | 14:53 | |
| *** dkranz has quit IRC | 14:55 | |
| Shrews | https://groups.google.com/forum/#!topic/gearman/osLOX3B-zy0 | 14:56 |
| *** boris-42 has quit IRC | 14:57 | |
| *** jjmb has joined #openstack-infra | 15:00 | |
| *** CaptTofu has quit IRC | 15:00 | |
| jeblair | Shrews: no, we're using gearman-java | 15:00 |
| *** dkranz has joined #openstack-infra | 15:01 | |
| *** boris-42 has joined #openstack-infra | 15:04 | |
| *** krtaylor has quit IRC | 15:05 | |
| *** dina_belova has joined #openstack-infra | 15:20 | |
| *** markmcclain has quit IRC | 15:27 | |
| *** vijendar has quit IRC | 15:31 | |
| fungi | Shrews: olaph: either of you planning to make it to the user group meeting at cisco on thursday afternoon? we should try to grab some drinks after, if so | 15:36 |
| *** dina_belova has quit IRC | 15:36 | |
| *** hashar has quit IRC | 15:36 | |
| *** rnirmal has joined #openstack-infra | 15:37 | |
| *** dina_belova has joined #openstack-infra | 15:37 | |
| *** sandywalsh has quit IRC | 15:37 | |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/zuul: Stats: break out job runtimes and counts by result https://review.openstack.org/38942 | 15:40 |
| olaph | fungi: up in the air at the moment - will know better later today | 15:41 |
| *** dkranz has quit IRC | 15:47 | |
| openstackgerrit | A change was merged to openstack-infra/zuul: Add option to validate job names https://review.openstack.org/38753 | 15:52 |
| *** mindjive1 is now known as mindjiver | 15:52 | |
| *** bnemec has joined #openstack-infra | 15:53 | |
| *** avtar has joined #openstack-infra | 15:53 | |
| *** sandywalsh has joined #openstack-infra | 15:53 | |
| *** pabelanger has quit IRC | 15:54 | |
| *** pabelanger has joined #openstack-infra | 15:55 | |
| *** UtahDave has joined #openstack-infra | 15:55 | |
| *** jjmb has quit IRC | 15:55 | |
| *** dkranz has joined #openstack-infra | 16:00 | |
| SergeyLukjanov | hi, could anyone please retrigger https://jenkins.openstack.org/job/savanna-tarball/16/ ? | 16:00 |
| SergeyLukjanov | (it fails because tag has been pushed right after the change merged and no tree was available in job) | 16:01 |
| Shrews | fungi: likely not making it this week | 16:02 |
| sdague | man there are a lot of changes waiting on a second +2 in requirements | 16:03 |
| *** vogxn has joined #openstack-infra | 16:04 | |
| *** toddmorey has joined #openstack-infra | 16:06 | |
| *** sarob has joined #openstack-infra | 16:06 | |
| toddmorey | I'm looking for a Jim Blair. I hear he frequents this bar. | 16:07 |
| jeblair | toddmorey: who wants to know? :) | 16:07 |
| jeblair | i mean yaaaay! it's toddmorey! | 16:07 |
| toddmorey | ha! | 16:08 |
| mordred | Shrews: yay! we chose correctly in deciding to not use gearman-java-service! | 16:08 |
| *** mrodden has quit IRC | 16:08 | |
| jeblair | mordred: or do you mean service-gearman-java? i can never remember. | 16:08 |
| mordred | jeblair: or the peoples-front-of-gearman ? | 16:09 |
| toddmorey | jeblair: So I'm about to send another note to Jonathan and Mark with a deadline. But aside from getting the royal blessing, I think we're pretty much ready. Then it's just a matter of getting Sebastian, et al familiar with our review / deployment process. | 16:09 |
| fungi | toddmorey: welcome back! | 16:09 |
| toddmorey | Mark says (with love) that I've been 'brainwashed' by the infrastructure team. I prefer to call it 'found enlightenment.' | 16:11 |
| jeblair | toddmorey: great. i think by now we're pretty good at getting new folks up to speed (i mean, 1200 people have done this before!) | 16:11 |
| mordred | toddmorey: well, getting Sebastian and friends up to speed should be fine ... | 16:11 |
| mordred | damn | 16:11 |
| mordred | jeblair beat me to the sentence | 16:11 |
| jeblair | :) | 16:12 |
| toddmorey | totally. I really meant that more on his side than ours. But, yeah, should be super easy onboarding. | 16:12 |
| jeblair | toddmorey: fungi makes this great kool-aid you should try. | 16:12 |
| *** danger_fo_away is now known as danger_fo | 16:13 | |
| fungi | it's grape! | 16:14 |
| toddmorey | My question: how far can we take it? Web projects they work on currently start in a basecamp / redmine-like system called Chili. But can we maybe follow the OpenStack approach from blueprint to production? | 16:15 |
| jeblair | toddmorey: all the way! we can definitely start using blueprints and lp bugs.... | 16:15 |
| *** rcleere has joined #openstack-infra | 16:16 | |
| jeblair | toddmorey: and i think this might also be a good early adopter for storyboard when it's ready | 16:16 |
| toddmorey | That's skipping ahead perhaps, first step is having the code in git/gerrit. But I like to have big goals. | 16:16 |
| toddmorey | jeblair: that's a great idea. | 16:16 |
| *** ^demon|zzz has quit IRC | 16:17 | |
| *** mgagne has quit IRC | 16:17 | |
| jeblair | toddmorey: it might be useful to be able to have stories (blueprints or bugs) for the website that target openstack release milestones (like havana) | 16:18 |
| jeblair | toddmorey: or bugs/blueprints that apply to say, both the website and the infrastructure project, which launchpad and storyboard will do. | 16:19 |
| toddmorey | My hope, too, with having content managed in git (similar to docs) is that we can have more involvement in keeping it synced with openstack releases. | 16:20 |
| *** vijendar has joined #openstack-infra | 16:20 | |
| *** mrodden has joined #openstack-infra | 16:21 | |
| annegentle | hola amigos! | 16:21 |
| jeblair | toddmorey: annegentle shows up whenever anyone says "docs" :) | 16:21 |
| *** mgagne has joined #openstack-infra | 16:22 | |
| clarkb | morning | 16:22 |
| *** dina_belova has quit IRC | 16:22 | |
| toddmorey | Essentially, if you noticed something you wanted to change on a page, you edit the markdown / Sphinx / html / whatever file and submit a pull request. No additional 'site admin' account required, just use your existing git user. | 16:22 |
| toddmorey | Democratize all the things, I say. | 16:22 |
| fungi | sounds idyllic | 16:23 |
| *** avtar has quit IRC | 16:23 | |
| *** SergeyLukjanov has quit IRC | 16:23 | |
| toddmorey | annegentle: hello! We need to talk | 16:23 |
| annegentle | toddmorey: yessir | 16:23 |
| toddmorey | annegentle: let's make that happen today. I think the storm has mostly blown over. | 16:23 |
| jeblair | toddmorey: +1 | 16:23 |
| annegentle | toddmorey: sweet | 16:23 |
| annegentle | toddmorey: I'm free after 1 or so | 16:23 |
| annegentle | toddmorey: pull requests not gerrit? Wowsa | 16:24 |
| annegentle | toddmorey: :) | 16:24 |
| toddmorey | the general rule is that no matter how long the deadline, 90% of submissions happen in the last two days. (in this case, summit speaking submissions.) | 16:24 |
| jeblair | annegentle: he meant gerrit. :) | 16:24 |
| annegentle | jeblair: :) | 16:24 |
| annegentle | jeblair: honestly I wouldn't mind experimenting with PRs for a doc, not sure which one. | 16:24 |
| toddmorey | annegentle: I did. :) Sorry. Have to retire the old lingo. | 16:24 |
| jeblair | we need to brainwash him a little more until he says "submit a review" instead of "submit a pull request" | 16:24 |
| annegentle | jeblair: but yeah so far I don't think walk up PRs will be all that helpful | 16:25 |
| toddmorey | annegentle: be gentle to the 'new guy' | 16:25 |
| annegentle | toddmorey: ha ha | 16:25 |
| annegentle | oh ok now I have to go to lunch. Wah. Ttyl | 16:25 |
| annegentle | toddmorey: I'll call you at 1:30 or so | 16:25 |
| toddmorey | sounds great! I'll be here | 16:25 |
| *** cp16net is now known as cp16net|away | 16:26 | |
| toddmorey | jeblair: you have time to see my master plan (very drafty)? It can wait… I know you are buried this week. | 16:26 |
| jeblair | toddmorey: yes i do! this takes precedence over anything else. :) | 16:27 |
| clarkb | fungi: mordred: jeblair: was there resolution on the pbr stuff in relation to jeepyb yesterday? | 16:27 |
| clarkb | I really want to get the new bug stuff up and going | 16:28 |
| fungi | clarkb: it sounds like we manually install any jeepyb dependencies on servers for now, because pbr doesn't support what our puppet module does yet, but jeblair doesn't want to work around it in puppet | 16:28 |
| fungi | clarkb: short answer is that pbr will not install dependencies globally when you run python setup.py install in the jeepyb source, but will when you run pip install . in the jeepyb source | 16:29 |
| jeblair | fungi: wait, this applies to more than just pbr? | 16:30 |
| fungi | jeblair: yes | 16:30 |
| jeblair | fungi: i thought the only issue was thit it didn't install pbr... but it doesn't install _anything_? | 16:30 |
| clarkb | can ww change puppet to use the install atring that works? | 16:30 |
| toddmorey | jeblair: very early, but I want to discuss with the dev team this week or next to get them thinking about it. https://dl.dropboxusercontent.com/u/2177046/site-framework.001.png | 16:30 |
| jeblair | mordred: so pbr breaks 'python setup.py install' ? | 16:30 |
| *** krtaylor has joined #openstack-infra | 16:30 | |
| fungi | jeblair: some of those dependencies disappear as runtime dependencies (pbr, d2to1) with the entrypoints rework in pbr 0.5.20 released last night | 16:30 |
| fungi | jeblair: but pycrypto (a transitive dependency of paramiko) also does not get globally installed | 16:31 |
| *** markmcclain has joined #openstack-infra | 16:32 | |
| jeblair | fungi: we count on 'python setup.py install' installing lots of python dependencies for other apps too (eg, gear for zuul, etc) | 16:33 |
| jeblair | fungi: is that broken too? | 16:33 |
| fungi | jeblair: it's hard to say as an anecdote because the places where we're installing them right now already probably had the appropriate dependencies installed previously before those projects got pbr-ized | 16:33 |
| jeblair | toddmorey: that looks really cool. fyi, there are some squid experts around here... | 16:34 |
| fungi | and one squid upstream maintainer hangs out in here, i believe | 16:34 |
| jeblair | toddmorey: we also might want to look at using rackspace cloud database ('trove') as the database component | 16:35 |
| mordred | jeblair: no. | 16:35 |
| mordred | pbr does not break python setup.py install | 16:35 |
| jeblair | toddmorey: then it's a bit more cloudy and we should be able to horizontally scale the app layer easier. | 16:35 |
| *** BobBall is now known as BobBall_Away | 16:35 | |
| mordred | I am looking in to the transitive issue | 16:35 |
| jeblair | mordred: okay, i'm clearly very confused as to what's going on. | 16:35 |
| clarkb | me too | 16:36 |
| mordred | yeah - there's at least 20 different interrelated things going on here and I have not gotten the chance to provide an actual full report on what's happening | 16:36 |
| mordred | yesterday was very busy | 16:36 |
| mordred | but I'm looking in to the pycrypto issue | 16:37 |
| *** cp16net|away is now known as cp16net | 16:37 | |
| *** pentameter has joined #openstack-infra | 16:38 | |
| jeblair | mordred: i do not want to overburden you with questions, so perhaps you could indicate whether it would be worthwhile for us to continue talking about specific solutions now, or whether we should instead wait for you to provide a more holistic picture? | 16:38 |
| mordred | I would like for you to wait for me to provide a larger picture | 16:38 |
| fungi | right, it sounds like where we left the missing pycrypto is that it installed into a virtualenv correctly but didn't get pulled in globally on a centos 6.4 python 2.6 system, so other possible interactions going on | 16:39 |
| mordred | yes. that is correct. so next on my list is to spin up a centos 6.4 system and work on global installs | 16:40 |
| mordred | to see if there is a weird version interaction of something that's causing oddness | 16:41 |
| *** CaptTofu has joined #openstack-infra | 16:41 | |
| clarkb | but this isnt a centos issue | 16:42 |
| mordred | no? | 16:42 |
| clarkb | review.o.o cant install jeepyb either | 16:42 |
| mordred | awesome | 16:42 |
| *** odyi has joined #openstack-infra | 16:42 | |
| mordred | so, it works in a virtualenv - so there is an interaction going on _somewhere_ in global installs | 16:42 |
| openstackgerrit | David Caro proposed a change to openstack-infra/jenkins-job-builder: Added more options to the Build Timeout plugin https://review.openstack.org/39329 | 16:43 |
| ekarlso- | jeblair: how do you use a local jenkins user for JJB in addition for launchpads ? | 16:43 |
| jeblair | ekarlso-: i do not do that | 16:43 |
| jeblair | ekarlso-: but once a user exists, a jenkins admin can get the api key for any user | 16:44 |
| *** ^demon|zzz has joined #openstack-infra | 16:44 | |
| *** ^demon|zzz has joined #openstack-infra | 16:44 | |
| toddmorey | jeblair: That's awesome. I know we did quite a bit of work with Squid @ rackspace. I think if we're smart about it, we could make most infrastructure blips almost undetectable. (Not that there will ever be any with this team at the wheel, though.) What drives me crazy about our current setup is that if there's any contention at all at the db level, the whole site can look to be offline, and that's really unnecessary. | 16:44 |
| jeblair | ekarlso-: so jjb uses that api key for a user (who may also be able to log in via launchpad) to do its work | 16:44 |
| mordred | clarkb: I have verified that python setup.py install works in a venv on centos6 | 16:45 |
| fungi | clarkb: oh, as for review{,-dev}.o.o i did manually install pbr last night on them to get things going again | 16:45 |
| jeblair | toddmorey: i think your design handles that well, minimizing the parts that talk to the db, making the rest very horizontally scalable and the whole thing cacheable | 16:46 |
| *** dina_belova has joined #openstack-infra | 16:46 | |
| mordred | I believe git.o.o is still broken right? | 16:46 |
| fungi | mordred: yes, or rather can be broken/unbroken simply by uninstalling packages with pip and then trying to setup.py install in the jeepyb source tree again | 16:46 |
| *** jjmb has joined #openstack-infra | 16:47 | |
| fungi | mordred: at the moment pip freeze is reporting pycrypto installed | 16:47 |
| mordred | fungi: I just uninstalled both | 16:48 |
| mordred | pycrypto and paramiko | 16:48 |
| fungi | mordred: i also haven't tried again since you released pbr 0.5.20 so maybe something there will fix it | 16:48 |
| mordred | and then I just ran python setup.py install | 16:48 |
| mordred | and it instlaled paramiko and pycrypto | 16:48 |
| mordred | :) | 16:48 |
| fungi | excellent! | 16:48 |
| mordred | ok. so - the thing that is doesn't do | 16:49 |
| mordred | which, if we consider it a bug I believe I konw how to fix | 16:49 |
| mordred | is that if I _ONLY_ uninstall pycrypto | 16:49 |
| fungi | is it possible that because gcc was originally not installed, paramiko got installed but pycrypto failed to compile/install, then later when i tried agai it thought the dependencies for paramiko were satisfied and didn't try to pull pycrypto in again? | 16:49 |
| mordred | and run python setup.py instlal | 16:49 |
| mordred | it does not attempt to install paramiko because it's up to date, and then it does not know to install pycrypto | 16:50 |
| fungi | ahh, yes. that sounds like a corner case we can probably deal with | 16:50 |
| mordred | I'm pretty sure we should consider that a bug | 16:50 |
| mordred | so - it's because of an explicit optimization I added | 16:50 |
| mordred | so I can totally take it out | 16:50 |
| mordred | because I think it's confusing, given this situation | 16:50 |
| *** CaptTofu has quit IRC | 16:51 | |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/zuul: Support multiple triggers https://review.openstack.org/39186 | 16:51 |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/zuul: Add a timer trigger (WIP) https://review.openstack.org/39187 | 16:51 |
| mordred | jeblair: ^^ does the above make sense, and/or do you dissent that I should remove the optimization? | 16:51 |
| mordred | jeblair: NEAT! | 16:51 |
| *** jjmb1 has joined #openstack-infra | 16:51 | |
| jeblair | mordred: reading | 16:51 |
| *** SergeyLukjanov has joined #openstack-infra | 16:51 | |
| fungi | that makes sense. so given that, and the removal of pbr/d2to1 as entrypoints wrapper runtime dependencies in pbr 0.5.20, and the decision not to support smooth transition from 0.5.17 to later pbr releases, seems like we're all buttoned up | 16:51 |
| mordred | I'm going to remove the optimization. I think it's more likely to screw people, and it's not that much of an optimization | 16:52 |
| *** emagana has quit IRC | 16:52 | |
| jeblair | mordred: what's the optimization you added? | 16:52 |
| fungi | and i can also abandon my jeepyb change for declaring pbr in requirements.txt now that 0.5.20 is out | 16:52 |
| mordred | it essentially just doesn't call pip if it detects that pip would be a no-op | 16:52 |
| jeblair | mordred: ah, but it can't really know that to a high degree of correctness without reimplementing a bunch of pip i guess. | 16:53 |
| mordred | right | 16:53 |
| jeblair | mordred: will it change sdist build time? | 16:53 |
| mordred | no | 16:54 |
| mordred | only install/develop | 16:54 |
| mordred | and not by much | 16:54 |
| mordred | it avoids a shell call to pip install at install time | 16:54 |
| *** jjmb has quit IRC | 16:54 | |
| mordred | which would return a "nothing to do" | 16:54 |
| jeblair | mordred: if tox starts doing 'setup develop', will the lack of this optimization make you said? | 16:54 |
| jeblair | sad | 16:54 |
| mordred | nope. because there's another optimization in tox around setup develop | 16:54 |
| mordred | which does not re-run develop if the egg-info is new enough | 16:55 |
| jeblair | ok. those are the cases where i think optimizing for time makes sense, i think normal usage of install/develop can handle being a teensy bit slower. | 16:55 |
| mordred | so tox won't run the develop command on ever tox invocation | 16:55 |
| mordred | ++ | 16:55 |
| *** pcrews has quit IRC | 16:55 | |
| *** datsun180b has quit IRC | 16:56 | |
| jeblair | mordred: no rush, but since you mentioned it, when you have a spare moment, "Support multiple triggers https://review.openstack.org/39186" is worth a look from you | 16:56 |
| *** sdague has quit IRC | 16:56 | |
| mordred | jeblair: yes! I'm very excited about it | 16:56 |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Remove missing_reuqires optimization https://review.openstack.org/39330 | 16:56 |
| jeblair | mordred: (not the second one, it's not remotely done yet) | 16:56 |
| *** tjones has joined #openstack-infra | 16:56 | |
| *** vogxn has quit IRC | 16:57 | |
| mordred | fungi: jeblair ^^ there is the patch that should fix the transitive issue | 16:57 |
| mordred | I have not tested it on the centos node yet | 16:57 |
| fungi | mordred: taking a look | 16:57 |
| *** emagana has joined #openstack-infra | 16:57 | |
| mordred | I'd like to grab it on git.o.o, install it, then reproduce our earlier failure | 16:57 |
| clarkb | fungi: so you upgraded pbr globally on the review hosts then reinstalled jeepyb?\ | 16:57 |
| mordred | and see if it fixes it | 16:57 |
| *** sarob_ has joined #openstack-infra | 16:57 | |
| *** tjones has left #openstack-infra | 16:58 | |
| fungi | clarkb: yep | 16:58 |
| jeblair | mordred: a change to pbr i feel i am competent to review! | 16:58 |
| openstackgerrit | A change was merged to openstack-infra/config: Add gcc package dependency to jeepyb module https://review.openstack.org/39151 | 16:59 |
| jeblair | pleia2: bug triage time? | 17:00 |
| *** sarob has quit IRC | 17:00 | |
| *** sarob__ has joined #openstack-infra | 17:00 | |
| clarkb | jeblair: yup, but pleia2 won't be joining us iirc | 17:01 |
| fungi | pleia2 is away from internets most of today, but prepped the etherpad for us in advance | 17:01 |
| jeblair | ah neat trick | 17:01 |
| *** ^demon|zzz is now known as ^d | 17:01 | |
| fungi | pretty awesome | 17:01 |
| *** sarob_ has quit IRC | 17:01 | |
| clarkb | https://etherpad.openstack.org/cibugreview-july2013 | 17:01 |
| jeblair | do we have any rax classic servers left? | 17:01 |
| jeblair | (other than old-wiki) | 17:02 |
| mordred | fungi: btw - sorry it took me so long for my brain to connect on what was happening there | 17:02 |
| *** gyee has joined #openstack-infra | 17:02 | |
| clarkb | jeblair: static moved, our oneiric hosts moved, I think we may be on rs nova now | 17:02 |
| fungi | mordred: no worries, i should have tried more combinations too, but it got late | 17:03 |
| clarkb | jog0: is there a bug related to testing client libs in tempest? we have a bug open against openstack-ci to test that which should be marked a duplicate | 17:03 |
| *** reed has joined #openstack-infra | 17:03 | |
| toddmorey | Anyone in the house stood up an OpenID environment before? | 17:03 |
| jeblair | oh we do have a few... | 17:03 |
| clarkb | jeblair: :/ | 17:04 |
| fungi | SergeyLukjanov: https://pypi.python.org/pypi/savanna/0.2.1 | 17:04 |
| clarkb | zaro: can you update https://bugs.launchpad.net/openstack-ci/+bug/1201035 and https://etherpad.openstack.org/cibugreview-july2013 with particulars about that job? | 17:05 |
| uvirtbot | Launchpad bug 1201035 in openstack-manuals "NullPointerException with docs maven build" [Medium,Confirmed] | 17:05 |
| clarkb | s/job/bug/ | 17:05 |
| fungi | SergeyLukjanov: from the original timestamps it looks like you pushed the tag for 0.2.1 before the commit it was tagging had gotten through testing and merged to the branch. i manually retriggered the tarball and then the pypi upload but you'll want to double-check it's correct | 17:05 |
| jeblair | toddmorey: a few years ago i used python-openid https://pypi.python.org/pypi/python-openid/ | 17:05 |
| jeblair | toddmorey: it provides a reasonable interface to being a provider as well as a consumer | 17:06 |
| SergeyLukjanov | fungi, thank you! | 17:06 |
| *** Ryan_Lane has joined #openstack-infra | 17:06 | |
| jeblair | mordred: paste.drizzle.org still points to unmaintained old openstack paste | 17:07 |
| jeblair | updated bug 1082787 | 17:08 |
| uvirtbot | Launchpad bug 1082787 in openstack-ci "Migrate remaining servers from rackspace classic to rackspace nova" [High,Triaged] https://launchpad.net/bugs/1082787 | 17:08 |
| clarkb | mordred: can you update 1203198 with info on why you think that happened in the first place? iirc you had some theories | 17:09 |
| jeblair | bug 1203198 | 17:09 |
| uvirtbot | Launchpad bug 1203198 in openstack-ci "Mirror jobs should not copy new packages if installation fails." [High,Triaged] https://launchpad.net/bugs/1203198 | 17:09 |
| *** psedlak has quit IRC | 17:10 | |
| *** sdague has joined #openstack-infra | 17:13 | |
| openstackgerrit | A change was merged to openstack-dev/hacking: Update I18N code style https://review.openstack.org/39217 | 17:14 |
| toddmorey | jeblair: thanks. My high-level questions are if it's possible or even advisable to store metadata (think IRC handle, tshirt size) about a user in an OpenID provider, or if its role should be strictly limited to authentication? My current thinking is that the metadata would remain in a separate data store, tied by a unique ID back to each account. | 17:17 |
| toddmorey | jeblair: secondly, API access tokens & authentication are considered an entirely separate problem, correct? | 17:18 |
| jeblair | toddmorey: it helps to think of openid as more of a protocol, so you can continue to store usermetadata however you want, but openid allows you to share it, along with an authentication assertion, with other services. | 17:19 |
| toddmorey | jeblair: yeah, protocol is a fantastic way to put it. | 17:19 |
| jeblair | toddmorey: so it makes sense for the foundation profile system to store that information, and then share it as appropriate with other apps via openid -- but also you can still have back-channel communication between apps if needed. | 17:20 |
| toddmorey | jeblair: folks are conflating member authentication with member data, and I'm trying to clear the air a bit. | 17:20 |
| jeblair | toddmorey: for example: http://imgur.com/KxZZ3X4 | 17:21 |
| jeblair | toddmorey: that's what i see when i log into jenkins using launchpad's openid auth... | 17:22 |
| jeblair | toddmorey: note that there are several bits of metadata there (including group membership); all of which were requested by jenkins | 17:22 |
| jog0 | clarkb: what is the bug? | 17:23 |
| toddmorey | jeblair: right, and that 'personal information metadata' is stored with the OpenID provider or with the app? (Though in this case, LP is playing the role of both I guess) | 17:23 |
| openstackgerrit | Clark Boylan proposed a change to openstack-infra/config: Start import process for clouddocs-maven-plugin. https://review.openstack.org/39337 | 17:24 |
| clarkb | jog0: https://bugs.launchpad.net/openstack-ci/+bug/939699 | 17:24 |
| uvirtbot | Launchpad bug 939699 in openstack-ci "client lib integration tests should hit multiple branches" [High,Triaged] | 17:24 |
| toddmorey | jeblair: Here's my swing at an overview, I welcome any ideas / corrections you may have to my thinking - https://dl.dropboxusercontent.com/u/2177046/MemberDataAndAuth.002.png | 17:24 |
| jeblair | toddmorey: with the app -- when someone hits the app endpoint with an openid request, the app would have some code to fetch whatever is needed from the db and supply the requested information. | 17:25 |
| jog0 | clarkb: ahh this is based on the clients should work across releases | 17:25 |
| toddmorey | jeblair: great, okay, that's very helpful and definitely fits the protocol model | 17:25 |
| jeblair | toddmorey: the canonical location would still be the db managed by the app; you're just exposing that information via a new channel: openid. | 17:25 |
| openstackgerrit | David Caro proposed a change to openstack-infra/jenkins-job-builder: Added pre-scm-buildstep wrapper https://review.openstack.org/39338 | 17:26 |
| jeblair | toddmorey: you might also consider exposing the same info via oauth or similar down the line as well. | 17:26 |
| jeblair | toddmorey: (just as an example) | 17:26 |
| jog0 | AFAIK it doesn't need any major support from tempest. The idea would be to take trunk client integration tests (some in tempest) and run them against all supported versions of openstack | 17:26 |
| jog0 | so when someone proposes a change to the client we know it doesn't break backwards compat ... | 17:27 |
| clarkb | jog0: I see. Could you update the openstack-ci with info about the current testing and what you think is necessary to meet the bugs goals? | 17:27 |
| jog0 | sure | 17:27 |
| clarkb | awesome, thank you | 17:28 |
| *** markmcclain has quit IRC | 17:28 | |
| clarkb | _TheDodd_: https://review.openstack.org/39337 used related-bug and it seems to have worked | 17:28 |
| jeblair | toddmorey: so one big question about the auth diagram is that i think a number of us are assuming that the member database would continue to store authentication information, eg, it would be an openid provider. | 17:28 |
| clarkb | _TheDodd_: we should probably do some rudimentary testing on the other headers too | 17:28 |
| _TheDodd_ | clarkb: Agreed. | 17:29 |
| jeblair | toddmorey: but the diagram makes it look like it would be an openid consumer. | 17:29 |
| _TheDodd_ | Is it reasonable to create some dummy bugs to simply run some testing against? | 17:29 |
| jeblair | toddmorey: i think we would want it to be a provider, both to help tie our different systems together, and to make it a good experience for new users who may not have an openid provider. | 17:30 |
| jeblair | toddmorey: (we could consider making it both a consumer and provider) | 17:30 |
| *** markmcclain has joined #openstack-infra | 17:30 | |
| *** sdague has quit IRC | 17:32 | |
| fungi | _TheDodd_: sure. we have at least one of those open already in openstack-ci | 17:32 |
| _TheDodd_ | Nice. | 17:32 |
| fungi | or maybe i marked it resolved | 17:32 |
| _TheDodd_ | lol | 17:32 |
| fungi | but can certainly be reopened too | 17:32 |
| jeblair | toddmorey: (this may need to be an email or phone conversation) | 17:32 |
| jog0 | clarkb: done | 17:32 |
| jeblair | bug 1082800 ! | 17:33 |
| uvirtbot | Launchpad bug 1082800 in openstack-ci "Add gearman support to zuul" [Medium,Fix released] https://launchpad.net/bugs/1082800 | 17:33 |
| fungi | _TheDodd_: or you can set up a separate lp org and open a bug under that if you want to avoid spamming us with updates | 17:33 |
| Alex_Gaynor | Anyone know what timezone dtroyer is in/what hours he keeps? | 17:33 |
| _TheDodd_ | fungi, haha | 17:33 |
| _TheDodd_ | a dangerous choice to leave in my hands, sir. | 17:33 |
| clarkb | jeblair: nice | 17:33 |
| clarkb | fungi: will that work? | 17:33 |
| jeblair | clarkb: it's nice if we can actually close some of these things occasionally! :) | 17:33 |
| fungi | clarkb: _TheDodd_: i meant separate project | 17:34 |
| jeblair | i don't mind spam for this | 17:34 |
| *** nati_ueno has joined #openstack-infra | 17:34 | |
| clarkb | neither do I | 17:34 |
| fungi | yeah, it's fine with me | 17:34 |
| jeblair | (indeed, if i were doing it, i would just spam us) :) | 17:34 |
| _TheDodd_ | fungi: OK, I will take care of that ASPA. | 17:35 |
| fungi | it threads, and i already know how to deal with large volumes of bugspam since i'm on every bugteam in openstack | 17:35 |
| fungi | (transitively through the vmt, so i can skim them for missed vulnerability reports) | 17:35 |
| *** bnemec has quit IRC | 17:35 | |
| clarkb | mordred: https://bugs.launchpad.net/openstack-ci/+bug/1010621 | 17:36 |
| uvirtbot | Launchpad bug 1010621 in openstack-ci "important servers should have backups" [Critical,In progress] | 17:36 |
| *** sdague has joined #openstack-infra | 17:36 | |
| *** bnemec has joined #openstack-infra | 17:36 | |
| _TheDodd_ | clarkb: The regular expression which adds hyperlinks to commit logs looks as though it could use some updating as well. | 17:38 |
| clarkb | _TheDodd_: oh yeah, that is in openstack-infra/config hidden somewhere /me finds it | 17:38 |
| _TheDodd_ | I'll track down the code and file a bug report as a wishlist .... | 17:38 |
| _TheDodd_ | ok | 17:38 |
| clarkb | _TheDodd_: http://logs.openstack.org/25/39325/2/check/gate-swift-devstack-vm-functional/194/ | 17:39 |
| clarkb | whoops wrong link | 17:39 |
| clarkb | _TheDodd_: https://github.com/openstack-infra/config/blob/master/modules/openstack_project/manifests/gerrit.pp#L133-L135 | 17:39 |
| clarkb | _TheDodd_: feel free to propose an update to the regex there | 17:39 |
| *** jjmb1 has quit IRC | 17:40 | |
| jeblair | i'd like to mark 917844 wontfix | 17:40 |
| jeblair | bug 917844 | 17:40 |
| uvirtbot | Launchpad bug 917844 in openstack-ci "Gerrit should error if an Approval is attempted without a +2 in code review" [Low,Triaged] https://launchpad.net/bugs/917844 | 17:40 |
| _TheDodd_ | clarkb: Sounds good. I'll put it on my shortlist. | 17:40 |
| clarkb | jeblair: wfm | 17:40 |
| jeblair | i don't think it's an operational problem at this point because zuul will not enqueue it. | 17:40 |
| clarkb | for https://bugs.launchpad.net/openstack-ci/+bug/1076481 we now have swift functional tests running as non voting against swift changes. Is that sufficient to close the bug or do we want to wait for them to be voting? | 17:41 |
| uvirtbot | Launchpad bug 1076481 in openstack-ci "Add Swift functional tests to full test run" [High,Triaged] | 17:41 |
| *** sdague has quit IRC | 17:41 | |
| jeblair | clarkb: are they passing? maybe we should make them voting | 17:41 |
| clarkb | jeblair: the one I checked on passed. I can ask notmyname and chmouel | 17:41 |
| clarkb | notmyname: chmouel: the new swift functional tests, can we make them voting tests? | 17:42 |
| fungi | jeblair: agreed, 917844 doesn't buy us anything to spend time on. it either is or isn't an upstream gerrit feature | 17:42 |
| *** sdague has joined #openstack-infra | 17:42 | |
| openstackgerrit | mark mcclain proposed a change to openstack-infra/reviewstats: update core list to reflect recent changes https://review.openstack.org/39343 | 17:42 |
| *** nayward has quit IRC | 17:42 | |
| clarkb | notmyname: chmouel: making them voting would mean adding the tests to the gate queue and removing the non voting property of the test | 17:43 |
| *** sarob__ has quit IRC | 17:43 | |
| *** avtar has joined #openstack-infra | 17:44 | |
| *** sarob has joined #openstack-infra | 17:44 | |
| *** vipul is now known as vipul-away | 17:46 | |
| openstackgerrit | A change was merged to openstack-infra/reviewstats: update core list to reflect recent changes https://review.openstack.org/39343 | 17:46 |
| *** vijendar has quit IRC | 17:46 | |
| openstackgerrit | A change was merged to openstack-infra/reviewstats: Use current username as default https://review.openstack.org/39150 | 17:47 |
| *** sarob has quit IRC | 17:48 | |
| clarkb | olaph: https://bugs.launchpad.net/openstack-ci/+bug/1172392 in the etherpad it says you can have a go at that one. Is that still the case? | 17:49 |
| uvirtbot | Launchpad bug 1172392 in openstack-ci "figure out how to do translation branches in transifex" [High,Triaged] | 17:50 |
| *** pcrews has joined #openstack-infra | 17:51 | |
| clarkb | fungi: https://bugs.launchpad.net/openstack-ci/+bug/1168455 what was the workaround? can we add it to the bug | 17:52 |
| uvirtbot | Launchpad bug 1168455 in openstack-ci "Etherpad export function always exports another document" [Medium,Triaged] | 17:52 |
| *** vijendar has joined #openstack-infra | 17:53 | |
| fungi | clarkb: oh! right... i think we had to manually craft the url by substituting the actual etherpad id for the null | 17:53 |
| fungi | i'll update the bug | 17:54 |
| jeblair | we should probably do a wiki style redirect and give in to etherpad's url demands. | 17:54 |
| _TheDodd_ | clarkb: I've got a lot going on today here at work, and I'm not sure that I'll have time to set up an entire project to test the bug functionality... | 17:54 |
| _TheDodd_ | clarkb: Is there a faster way to get the necessary tests done, possibly? | 17:55 |
| fungi | _TheDodd_: no need. sounds like we're all fine with being spammed | 17:55 |
| _TheDodd_ | Okay, where can I find the dummy bug to being spamming you guys? | 17:56 |
| _TheDodd_ | * to begin spamming | 17:56 |
| fungi | _TheDodd_: just a sec, i'll dig it up | 17:56 |
| jeblair | you could reopen 1037327 | 17:57 |
| jeblair | bug 1037327 | 17:57 |
| uvirtbot | Launchpad bug 1037327 in openstack-ci "Test bug for gerrit bug notifications" [Undecided,Fix released] https://launchpad.net/bugs/1037327 | 17:57 |
| fungi | yep, that's the one i was searching for. lp was being sluggish to search | 17:57 |
| clarkb | _TheDodd_: https://bugs.launchpad.net/openstack-ci/+bug/809479 | 17:57 |
| uvirtbot | Launchpad bug 809479 in openstack-ci "gerrit integration test" [Undecided,Invalid] | 17:57 |
| clarkb | heh we have more than one | 17:57 |
| fungi | apparently | 17:57 |
| *** sarob has joined #openstack-infra | 17:59 | |
| jeblair | i've triaged the new bugs | 18:00 |
| clarkb | jeblair: https://bugs.launchpad.net/openstack-ci/+bug/1172407 and https://bugs.launchpad.net/openstack-ci/+bug/1082794 can we close the second and use the first to track symmetric testing? | 18:00 |
| uvirtbot | Launchpad bug 1172407 in openstack-ci "forward test proposed changes to old branch with grenade" [Medium,Triaged] | 18:00 |
| clarkb | jeblair: or should we mark one as a dup? | 18:00 |
| jeblair | except bug 1192039 | 18:00 |
| uvirtbot | Launchpad bug 1192039 in openstack-ci "Merge release tags back into master" [Undecided,New] https://launchpad.net/bugs/1192039 | 18:00 |
| jeblair | and bug 1206073 | 18:00 |
| uvirtbot | Launchpad bug 1206073 in openstack-ci "cinder volume fails to report in if jsonschema doesn't match what it was compiled with" [Undecided,New] https://launchpad.net/bugs/1206073 | 18:00 |
| jeblair | mordred: can you look at 1192039 ? | 18:00 |
| *** sarob has quit IRC | 18:01 | |
| jeblair | clarkb: yeah, let's mark the second as a dup | 18:01 |
| clarkb | danke | 18:01 |
| clarkb | fungi: https://bugs.launchpad.net/openstack-ci/+bug/1172425 seems up your alley | 18:01 |
| uvirtbot | Launchpad bug 1172425 in openstack-ci "add python3 tox env with test whitelist" [Low,Triaged] | 18:01 |
| *** sarob has joined #openstack-infra | 18:01 | |
| fungi | i've been trying to stay on top of the triage as they land, but have missed at least a few | 18:01 |
| fungi | clarkb: nabbing that now | 18:01 |
| clarkb | jeblair: dup marked | 18:03 |
| *** koolhead17 has joined #openstack-infra | 18:03 | |
| *** sarob has quit IRC | 18:04 | |
| fungi | clarkb: do we have any examples of whitelisted tests in tox.ini? i'm thinking 1172425 isn't a ci bug, but rather a bug to be addressed in each project being tested | 18:04 |
| *** sarob has joined #openstack-infra | 18:04 | |
| clarkb | fungi: ya, I think the bug that was submitted against openstack-ci is there so that we know this is a thing other people want to do | 18:06 |
| clarkb | fungi: now we have slaves with python 3.3 so our end is done | 18:07 |
| clarkb | jog0: ^ btw you can run hacking py3k tests now | 18:07 |
| fungi | well, once we get more core reviews on https://review.openstack.org/37426 at any rate | 18:07 |
| fungi | those will be our first projects gated on py3k | 18:07 |
| * fungi looks to see if hacking py33 works | 18:08 | |
| clarkb | fungi: if we approve 37426 will we prevent merges in those projects until they pass py33 tests? | 18:09 |
| clarkb | oh I see they are only check tests | 18:09 |
| fungi | clarkb: they could in fact be made gating now. they passed when i tried them on the new slaves last night | 18:09 |
| clarkb | no that is layout dev. I confuse myself | 18:10 |
| clarkb | fungi: cool. I am just going to approve and see what happens. The fast zuul config updates makes it relatively safe | 18:10 |
| * fungi nods. check-only on zuul-dev and gating on prod | 18:10 | |
| jeblair | clarkb: +1 | 18:10 |
| fungi | clarkb: awesome! then in the meeting i can say we're actually gating projects on python 3.3 | 18:10 |
| clarkb | :) | 18:10 |
| fungi | rather than plug for reviews | 18:11 |
| jog0 | clarkb: nice | 18:12 |
| fungi | jog0: trying py33 on hacking now... https://jenkins.openstack.org/job/fungi-node-tester/42/console | 18:12 |
| jog0 | fungi: cool trying it on my local box too | 18:13 |
| jog0 | last time i tried I think pbr broke | 18:13 |
| jog0 | but that was a while ago | 18:13 |
| fungi | jog0: pbr is now py33 clean | 18:13 |
| fungi | jog0: as is oslo.config | 18:13 |
| jog0 | nice | 18:14 |
| fungi | jog0: in mere moments, they'll be officially gated on it. if hacking works, i'll throw up a quick review to add it too | 18:14 |
| jog0 | so pbr will be the first to py33 gate? | 18:14 |
| fungi | pbr and oslo.config both | 18:14 |
| jog0 | nice | 18:14 |
| sdague | mordred: so requirements repo, before trying to make update.py do the right thing, I feel like we need to build some testing into it, otherwise this is going to be a bug fix chace | 18:15 |
| jog0 | you guys should send out a ML email, as this is a milestone | 18:15 |
| fungi | jog0: i guess i can do that this afternoon | 18:15 |
| * fungi doesn't count his chickens until they hatch | 18:15 | |
| jog0 | fungi: ohh nice the unit test are failing for py33 in hacking. exactly what we wanted | 18:15 |
| clarkb | fungi: jeblair: I am updating zuul and git review to say fix released on https://bugs.launchpad.net/zuul/+bug/1172444 I don't seem to have permsto do this for gear but gear appears to be using flake8 too | 18:15 |
| jeblair | fungi: ++email | 18:15 |
| uvirtbot | Launchpad bug 1172444 in python-heatclient "migrate all projects to flake8" [Medium,Fix committed] | 18:15 |
| *** ^d has quit IRC | 18:16 | |
| *** ^d has joined #openstack-infra | 18:16 | |
| *** ^d has joined #openstack-infra | 18:16 | |
| fungi | jog0: yep, looks like it bailed on print statement vs. print() function | 18:16 |
| jeblair | clarkb: that's fascinating that it could be added to gear which has no bugtracker configured in lp... | 18:16 |
| jeblair | i wonder if it used to, but then got fubar during the rename | 18:17 |
| clarkb | jeblair: o_O | 18:17 |
| fungi | jog0: oh, it needs newer testr | 18:17 |
| fungi | jog0: and ConfigParser vs. configparser | 18:18 |
| *** sarob has quit IRC | 18:18 | |
| jog0 | fungi: ConfigParser is inside of hacking | 18:19 |
| *** sarob has joined #openstack-infra | 18:19 | |
| jog0 | if that is the bug we see then py33 is gateable | 18:19 |
| jog0 | (once we fix things) | 18:19 |
| fungi | yeah, that's a minor and easily addressed one | 18:19 |
| openstackgerrit | A change was merged to openstack-infra/config: Enable python33 gate/check for pbr and oslo.config https://review.openstack.org/37426 | 18:19 |
| fungi | i'm looking to see if subunit2html.py needs help | 18:20 |
| clarkb | fungi: it may | 18:20 |
| mordred | sdague: I agree | 18:20 |
| jeblair | clarkb: gear-bugs created, made supervisor, and i marked that bug as fix released | 18:20 |
| mordred | jeblair, clarkb: I have seen all of your queries from above - sorry, I got sucked into 1.5 hours of phone | 18:20 |
| clarkb | jeblair: ty | 18:20 |
| mordred | will start slogging through now | 18:20 |
| jeblair | mordred: understood; i tried to keep it minimal. | 18:21 |
| *** toddmorey has quit IRC | 18:21 | |
| fungi | clarkb: looks like its subunut2html.py getting called from tox in py33 which is using the print statement, so it'll need cleaning. review coming | 18:21 |
| sdague | mordred: so you know what gets interesting..... trying to call tox in requirements | 18:21 |
| mordred | hah. | 18:22 |
| sdague | because requirements.txt isn't the requirements for the repo | 18:22 |
| mordred | sdague: so, make the tox.ini in reuqirements not reference requirements.txt | 18:22 |
| mordred | just put them directly into the deps list | 18:22 |
| sdague | it doesn't | 18:22 |
| mordred | oh | 18:22 |
| sdague | but I can't use the testr setup.py | 18:22 |
| sdague | because pbr magic | 18:22 |
| mordred | did we put a setup.py in there? | 18:22 |
| clarkb | fungi: https://bugs.launchpad.net/openstack-ci/+bug/1200215 I updated that with a little more info on the fails around git revert and our commit message hokk | 18:22 |
| uvirtbot | Launchpad bug 1200215 in openstack-ci "Gerrit revert does not include Change-Id in commit message" [Medium,Triaged] | 18:22 |
| clarkb | *hook | 18:22 |
| sdague | mordred: yes | 18:23 |
| mordred | sdague: ahhhhh | 18:23 |
| *** afazekas has quit IRC | 18:23 | |
| sdague | so if you have an unwind for me, that's good as well | 18:23 |
| fungi | clarkb: note thought that it was about gerrit's revert button. also i started playing around with hooking that to other places besides commit-msg but it can have unintended side effects | 18:23 |
| *** sarob has quit IRC | 18:23 | |
| fungi | s/thouht/though/ | 18:24 |
| clarkb | fungi: ya, I wonder if gerrit is using the same commit hook on their end | 18:24 |
| clarkb | ooh unintended side effects | 18:24 |
| clarkb | fungi: does the commit message hook just not fire on a revert? or does that hook itself check what called it and exit early if in a revert? | 18:25 |
| _TheDodd_ | fungi: jeblair: clarkb: Gents, what should I do for the test? Should I commit a README for testing, and then remove it in another test? I don't want to do anything too wild and end up breaking something. Maybe I should just delete the whole repository and commit that :D | 18:25 |
| *** nicedice has joined #openstack-infra | 18:25 | |
| clarkb | _TheDodd_: I think if you submit several patchsets against a change that use the various regexes and make sure that LP updates that is probably sufficient | 18:26 |
| _TheDodd_ | It looks like that is what was done in earlier tests ( https://review-dev.openstack.org/#/c/94/ ) | 18:26 |
| fungi | clarkb: revert, merge, cherry-pick et al do not use the same higher-level plumbing which calls hooks when editing commit messages, and hooking them directly can potentially cause us to run the script at undesirable times. i haven't completely fleshed out the counterargument yet | 18:26 |
| *** nicedice has joined #openstack-infra | 18:26 | |
| _TheDodd_ | ok | 18:26 |
| clarkb | fungi: interesting | 18:26 |
| *** vipul-away is now known as vipul | 18:27 | |
| jeblair | in order to test reacting to merges, we'll need to actually merge something | 18:27 |
| jeblair | review-dev would be ideal for that | 18:27 |
| jeblair | (we should really manage that with jeepyb) | 18:27 |
| fungi | clarkb: but point being the other hooks fire when those operations happen rather than when the commit message is being edited, so you would for example prepopulate the commit message with a change-id | 18:27 |
| mordred | sdague: I think I have an idea... | 18:28 |
| *** atiwari has joined #openstack-infra | 18:28 | |
| clarkb | jeblair: my wiki doc update change can be used to test reacting to merges | 18:28 |
| clarkb | jeblair: or some change along those lines | 18:28 |
| jeblair | ok, that's probably easiest then | 18:28 |
| mordred | sdague: put SKIP_PIP_INSTALL=1 into the environment section | 18:29 |
| clarkb | I will need to update the commit message but that is a low impact change that should be safe to push through | 18:29 |
| mordred | sdague: and then put direct deps into the tox.ini file | 18:29 |
| sdague | I need pip install | 18:29 |
| sdague | oh... hmmm | 18:29 |
| sdague | you got a sample example? | 18:29 |
| mordred | sdague: yah. one sec | 18:29 |
| sdague | I think if I create the py27 directly in tox.ini that would work as well | 18:29 |
| sdague | basically just need testr and testtools | 18:29 |
| openstackgerrit | Jeremy Stanley proposed a change to openstack-infra/config: Make subunit2htm.py Py3k compatible. https://review.openstack.org/39356 | 18:30 |
| mordred | sdague: http://paste.openstack.org/show/42655 | 18:31 |
| *** sarob has joined #openstack-infra | 18:31 | |
| openstackgerrit | Clark Boylan proposed a change to openstack-infra/config: Add Wiki upgrade process documentation. https://review.openstack.org/39137 | 18:32 |
| openstackgerrit | Jeremy Stanley proposed a change to openstack-infra/config: Make subunit2html.py Py3k compatible. https://review.openstack.org/39356 | 18:32 |
| clarkb | jeblair: _TheDodd_ ^ that change has been updated to use Closes-Bug so that we can test the on merge behaviour | 18:32 |
| mordred | sdague: also, woot, glance landed the rename requirements change | 18:33 |
| sdague | mordred: trying | 18:33 |
| *** melwitt has joined #openstack-infra | 18:33 | |
| sdague | mordred: cool... how long was that in the queue? | 18:33 |
| mordred | sdague: since may I think? | 18:33 |
| clarkb | wow | 18:33 |
| sdague | mordred: yeh, that didn't help | 18:33 |
| clarkb | fungi: re https://review.openstack.org/#/c/39356/2 when is the from __future__ import necessary? | 18:34 |
| clarkb | is that just for python < 2.6? | 18:34 |
| sdague | clarkb: <= 2.6 | 18:35 |
| sdague | i'm pretty sure | 18:35 |
| clarkb | fungi: we will probably need that import then | 18:35 |
| openstackgerrit | Sean Dague proposed a change to openstack/requirements: work in progress... starting to add testing https://review.openstack.org/39358 | 18:36 |
| sdague | mordred: can you look at that ^^^^ | 18:36 |
| *** koolhead17 has quit IRC | 18:37 | |
| fungi | clarkb: oh, from __future__ import print or what? | 18:37 |
| mordred | sdague: I'm excited | 18:37 |
| * fungi tries | 18:37 | |
| mordred | fungi: from __future__ import print_function | 18:37 |
| sdague | mordred: yeh, well I need your help on tox-fu | 18:38 |
| sdague | once I get past that I can get the functionality work | 18:38 |
| sdague | working | 18:38 |
| clarkb | jeblair: https://bugs.launchpad.net/openstack-ci/+bug/1103082 do you think that is related to the Jenkins <-> d-g stuff that was getting out of sync? perhaps we can close that bug? | 18:38 |
| uvirtbot | Launchpad bug 1103082 in openstack-ci "Build failed with " You are already running a stack.sh session"" [Low,Triaged] | 18:38 |
| clarkb | jeblair: I can mark it as a dup which will pick up the released status | 18:39 |
| jeblair | clarkb: it's related to d-g fragility. i mean we could close it. | 18:39 |
| fungi | clarkb: sdague: mordred: print() works on my python2.6 | 18:40 |
| fungi | without __future__.print_function | 18:40 |
| sdague | oh... interesting | 18:40 |
| sdague | then why did we land it in nova? | 18:40 |
| sdague | I have no 2.6 anywhere to test any more | 18:40 |
| fungi | "New in version 2.6." | 18:40 |
| fungi | http://docs.python.org/2/library/functions.html#print | 18:40 |
| jeblair | for python 2.5 compat? :) | 18:41 |
| mordred | sdague: it looks great! | 18:41 |
| fungi | 2.5 would need it, yes | 18:41 |
| mordred | oh - great. we should remove the future print_function stuff - and add a hacking check for it :) | 18:41 |
| sdague | mordred: tox doesn't run for me on that tree | 18:41 |
| mordred | sdague: ok. grabbing | 18:41 |
| sdague | that's what I need the help with :) | 18:41 |
| pabelanger | GheRivero: thanks again for the sponsor | 18:42 |
| GheRivero | pabelanger: :) | 18:43 |
| clarkb | reed: https://bugs.launchpad.net/openstack-ci/+bug/1168417 can you provide me with what you think is the proper logo to use? I can update the github org with that image | 18:43 |
| uvirtbot | Launchpad bug 1168417 in openstack-ci "https://github.com/openstack displays a low-res OpenStack logo" [Low,Triaged] | 18:43 |
| reed | clarkb, http://www.openstack.org/brand/openstack-logo/ | 18:44 |
| * clarkb tries with the small png | 18:45 | |
| dstufft | print() isn't a function on 2.x | 18:45 |
| dstufft | it's doing print () | 18:45 |
| mordred | haha. nice | 18:46 |
| dstufft | it just so happens the syntax works, but you can't use keyword arguments or anything | 18:46 |
| fungi | cute way of doing it | 18:46 |
| mordred | sdague: sigh | 18:47 |
| mordred | sdague: SKIP_PIP_INSTALL=1 tox | 18:47 |
| mordred | sdague: will work for you | 18:47 |
| clarkb | dstufft: so we should use the import everywhere to be fully compatible? | 18:47 |
| mordred | sdague: I can add an option to pbr to allow us to set that option in setup.cfg | 18:47 |
| mordred | sdague: or it'll work in tox 1.5.1 once it's released | 18:47 |
| fungi | confirmed, type(print) doesn't return a function on 2.x without print_function, unlike 3.x | 18:47 |
| sdague | mordred: pbr might be the best route, need to work with existing tox | 18:48 |
| dstufft | clarkb: *shrug*, depends on what python versions you support and what you need from print. If you only do simple print <foo> then it doesn't really matter, if you need to change encodings or redirect the output stream or things like that the import is the better otpion | 18:48 |
| mordred | sdague: k. I'll get you a patch | 18:48 |
| *** ^d_ has joined #openstack-infra | 18:48 | |
| fungi | however, right now we don't have projects using exclusively 3.x, so we already test their syntax is valid under 2.x | 18:48 |
| mordred | fungi: ah - I think what dstufft is saying is that if we keep the import print_function, then we can start using new features of print from 3.x in 2.x | 18:49 |
| clarkb | I think the openstack gravatar is under the not.mn domain | 18:49 |
| mordred | clarkb: of course it is | 18:49 |
| clarkb | mordred: jeblair ^ are you aware of the history there and whether or not we are able to update it? | 18:49 |
| clarkb | notmyname: ^ | 18:49 |
| mordred | clarkb: ther eis an openstack gravatar? | 18:49 |
| mordred | clarkb: what is it used for? | 18:49 |
| clarkb | mordred: yes that is where the logo on github comes from | 18:49 |
| mordred | ah. how funny | 18:49 |
| fungi | mordred: agreed. question of whether we take advantage of py3kisms in 2.x or stick with pure 2.x which works in 3.x | 18:49 |
| jeblair | clarkb: just change the address to something else and make a gravatar for it | 18:49 |
| dstufft | mordred: yes, that gives you 3.x stuff. If you don't need 3.x stuff you can still do print(…) in the simple case to keep the source compat without using the import | 18:49 |
| clarkb | jeblair: ok | 18:49 |
| *** ^d has quit IRC | 18:50 | |
| *** ^d_ is now known as ^d | 18:50 | |
| *** ^d has joined #openstack-infra | 18:50 | |
| clarkb | I am going to use review@openstack.org | 18:50 |
| jeblair | clarkb: +1 | 18:50 |
| fungi | wfm | 18:50 |
| clarkb | ugh have to create an account with wordpress.com connect so that I can have a gravatar? | 18:52 |
| jeblair | clarkb: this bug was not tagged 'low-hanging-fruit'. :) | 18:52 |
| clarkb | ya I am going to punt this now. I figured I could do it in a minute and close the bug but apparently not | 18:52 |
| sdague | mordred: SKIP_PIP_INSTALL=1 tox doesn't work for me | 18:52 |
| mordred | k. lemme make you apatch | 18:53 |
| mordred | sdague: also, do you have latest pbr installed on your machine? | 18:53 |
| sdague | probably not | 18:53 |
| fungi | ahh, i see. rereading the stdlib functions section on print() it's the __future__.prit_function which is new in 2.6, and allows for parameterization, specific separator/terminator, output redirection, et cetera | 18:53 |
| sdague | isn't that what tox is for? | 18:53 |
| sdague | so I don't have to constantly update my env :) | 18:54 |
| mordred | yes - once 1.5.1 is out | 18:54 |
| mordred | pbr is a setup_Requires - you need itto run the install to install into tox - the way it's currently written | 18:54 |
| mordred | chicken/egg | 18:54 |
| mordred | sdague: is there a pbr egg in your local dir? | 18:54 |
| sdague | yes | 18:54 |
| mordred | delete it. try again | 18:54 |
| sdague | and d2to1 | 18:55 |
| mordred | delete both | 18:55 |
| mordred | jeblair, fungi: ok - I tested the remove optmization patch | 18:55 |
| fungi | mordred: yay? | 18:55 |
| mordred | jeblair, fungi and it did what we want it to do - I'm going to land it | 18:55 |
| fungi | yay! | 18:55 |
| *** koolhead17 has joined #openstack-infra | 18:56 | |
| *** koolhead17 has quit IRC | 18:56 | |
| *** koolhead17 has joined #openstack-infra | 18:56 | |
| clarkb | ++ | 18:56 |
| sdague | mordred: yeh, still no dice | 18:56 |
| clarkb | sdague: mtreinish: https://bugs.launchpad.net/openstack-ci/+bug/1172479 isn't that in progress? Can it be assigned to someone? | 18:56 |
| uvirtbot | Launchpad bug 1172479 in openstack-ci "run tempest stress test as a periodic job" [Low,Triaged] | 18:56 |
| sdague | clarkb: assign to dkranz | 18:57 |
| *** toddmorey has joined #openstack-infra | 18:57 | |
| mtreinish | clarkb: that's done | 18:57 |
| mtreinish | https://jenkins.openstack.org/job/periodic-tempest-devstack-vm-stress/ | 18:57 |
| clarkb | I have assigned it to dkranz. mtreinish I can mark it as fix released? | 18:57 |
| clarkb | cool | 18:57 |
| mtreinish | clarkb: yeah that should be fine | 18:57 |
| sdague | mordred: do I need this patch_tox_venv.py? | 18:58 |
| *** markmcclain has quit IRC | 18:58 | |
| sdague | that cinder and other projects seem to use | 18:59 |
| mordred | no | 18:59 |
| sdague | hmmmm.... | 18:59 |
| mordred | it's for dealing with redhat's breakage of eventlet | 18:59 |
| sdague | ok | 18:59 |
| mordred | by patching python 2.6 with features from 2.7 | 18:59 |
| mordred | thanks redhat | 18:59 |
| fungi | mordred: didn't i hear a few weeks ago that they finally fixed that? | 19:00 |
| mordred | fungi: did they? | 19:00 |
| mordred | russellb: ^^ ? | 19:00 |
| *** toddmorey has quit IRC | 19:00 | |
| fungi | either latest python package on rhel or latest upstream eventlet (i forget which) supposedly ironed it out | 19:00 |
| russellb | mordred: don't know ... | 19:00 |
| fungi | i want to say someone mentioned it on the -dev list last month, but i'd need to go spelunking to find where i saw it now | 19:01 |
| jeblair | meeting time! | 19:01 |
| clarkb | mordred: s/2.7/3.X/ | 19:01 |
| clarkb | mordred: even more exciting | 19:01 |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Add skip_pip_install to setup.cfg https://review.openstack.org/39363 | 19:02 |
| mordred | sdague: ^^ | 19:02 |
| *** vipul is now known as vipul-away | 19:03 | |
| *** vipul-away is now known as vipul | 19:03 | |
| sdague | mordred: ok, cool, what's required to get that into something I can use here? | 19:04 |
| mordred | land it, have me cut a new pbr | 19:04 |
| *** markmcclain has joined #openstack-infra | 19:04 | |
| sdague | jeblair, fungi you guys want to review it? :) pretty please. (/me really wants to deal with this global requirements stuff once and for all) | 19:05 |
| *** toddmorey has joined #openstack-infra | 19:06 | |
| mordred | sdague: they're in openstack infra meeting - but we can hound them after it | 19:06 |
| sdague | ok | 19:06 |
| *** rnirmal has quit IRC | 19:07 | |
| *** rnirmal has joined #openstack-infra | 19:07 | |
| mrodden | anyone else run into crazy weird issues with the new tools/config/check_uptodate.sh ? | 19:08 |
| mrodden | i got this | 19:08 |
| mrodden | http://paste.openstack.org/show/42659/ | 19:09 |
| mrodden | so the one it generates has an extra blank line from when i updated? | 19:09 |
| *** rnirmal has quit IRC | 19:09 | |
| *** rnirmal has joined #openstack-infra | 19:10 | |
| mrodden | i like the concept but at the moment that check seems to be pretty flaky | 19:10 |
| *** changbl has quit IRC | 19:12 | |
| *** avtar has quit IRC | 19:12 | |
| *** sarob has quit IRC | 19:14 | |
| *** sarob has joined #openstack-infra | 19:15 | |
| *** vipul is now known as vipul-away | 19:18 | |
| *** sarob has quit IRC | 19:20 | |
| *** changbl has joined #openstack-infra | 19:21 | |
| *** gyee has quit IRC | 19:24 | |
| *** toddmorey has quit IRC | 19:24 | |
| clarkb | sdague: is that you hitting logstash? I am so happy when I see other people using it | 19:25 |
| sdague | nope, not I | 19:25 |
| clarkb | interesting, I wonder who is using it | 19:26 |
| clarkb | mrodden: no idea, folks in #openstack-nova may have more info | 19:26 |
| ttx | reed, rockstar: the openstack@lists.o.o list looks pretty slow, even by LP standards | 19:31 |
| ttx | Is it because it's giganormous ? | 19:31 |
| rockstar | ttx, I don't know. I wondered about that too. | 19:31 |
| ttx | Looks like it takes 3 hours to deliver a message | 19:31 |
| rockstar | Whoa. That seems crazy. | 19:32 |
| ttx | Date: Tue, 30 Jul 2013 16:17:05 +0200 | 19:32 |
| rockstar | ttx, I can take a look at it in a bit. | 19:32 |
| ttx | Received: Tue, 30 Jul 2013 10:21:40 -0700 (PDT) | 19:32 |
| zul | hey can we turn python3 cinderclient as well? | 19:32 |
| pentameter | Youngsters these days are so spoiled. | 19:32 |
| clarkb | zul: yes, did you want us to write that change for you? | 19:32 |
| ttx | pentameter: in the old days i was the only one on the internet so my MLs were delivered pretty fast. | 19:32 |
| pentameter | Internet? Dial up UCP & bang paths! | 19:33 |
| zul | clarkb: well if someone points me in the right direction then i could do that myself | 19:33 |
| ttx | pentameter: but thanks for calling me young. I appreciate that. | 19:33 |
| zul | ttx: he doesnt know you too well does he? ;) | 19:33 |
| ttx | zul: dude, I'm not THAT old. | 19:34 |
| clarkb | zul: openstack-infra/config then in modules/openstack_project/files/jenkins_job_builder/config/projects.yaml you need to add the gate-{name}-python33 template to the cinderclient project | 19:34 |
| zul | clarkb: cool ill get'er done then | 19:34 |
| clarkb | zul: then in modules/openstack_project/files/zuul/layout.yaml add gate-python-cinderclient-python33 to the check and gate queues | 19:34 |
| clarkb | for cinderclient | 19:34 |
| zul | gotcha | 19:35 |
| zul | ttx: sure sure :) | 19:35 |
| fungi | zul: we turned it on for pbr and oslo in https://review.openstack.org/37426 | 19:35 |
| clarkb | fungi: I just realized ^ did not change JJB | 19:36 |
| clarkb | fungi: was that done in a different change or did we miss that? | 19:36 |
| zul | fungi: cool congrats | 19:36 |
| fungi | clarkb: it depended on that change, which went in earlier | 19:36 |
| zul | fungi: i might do a refresh in the python3 ppa | 19:37 |
| fungi | clarkb: zul: https://review.openstack.org/37326 | 19:37 |
| fungi | almost the same change number, separated by exactly100 | 19:37 |
| *** dina_bel_ has joined #openstack-infra | 19:37 | |
| fungi | so that laid the jjb groundwork to get the builder and template implemented | 19:38 |
| *** markmcclain has quit IRC | 19:39 | |
| fungi | shouldn't need to be repeated for the change zul wants though | 19:39 |
| zul | fungi: ack | 19:39 |
| *** toddmorey has joined #openstack-infra | 19:39 | |
| *** dina_belova has quit IRC | 19:39 | |
| clarkb | fungi: the python33 job is not part of the python jobs job group | 19:40 |
| clarkb | fungi: so it need to be added explicitly to individual projects | 19:40 |
| fungi | clarkb: right | 19:40 |
| fungi | clarkb: oh, right, it's not added to pbr and oslo.config | 19:40 |
| fungi | fixing | 19:40 |
| openstackgerrit | Sean Dague proposed a change to openstack/requirements: add update.py testing to requirements https://review.openstack.org/39358 | 19:41 |
| openstackgerrit | Jeremy Stanley proposed a change to openstack-infra/config: Create python33 gate jobs for pbr and oslo.config. https://review.openstack.org/39365 | 19:44 |
| *** nati_uen_ has joined #openstack-infra | 19:44 | |
| sdague | so I think that's about as far as I can get before the pbr fix / release | 19:45 |
| clarkb | fungi: you need to do that same for pbr I htink | 19:45 |
| fungi | thought i did | 19:45 |
| *** nati_uen_ has quit IRC | 19:46 | |
| *** markmcclain has joined #openstack-infra | 19:47 | |
| *** nati_ueno has quit IRC | 19:47 | |
| *** avtar has joined #openstack-infra | 19:47 | |
| *** nati_ueno has joined #openstack-infra | 19:47 | |
| *** dina_bel_ has quit IRC | 19:48 | |
| *** sarob has joined #openstack-infra | 19:49 | |
| *** avtar has quit IRC | 19:49 | |
| *** avtar has joined #openstack-infra | 19:50 | |
| *** markmcclain has quit IRC | 19:51 | |
| clarkb | pleia2: bug day is going. I don't think we have managed to get through all of them yet though | 19:51 |
| pleia2 | great :) | 19:51 |
| pleia2 | I'm here for the next hour or so | 19:51 |
| *** vipul-away is now known as vipul | 19:53 | |
| *** afazekas has joined #openstack-infra | 19:55 | |
| mtreinish | clarkb,fungi,jeblair: any idea what happened here: http://logs.openstack.org/57/39357/1/check/gate-tempest-devstack-vm-postgres-full/37139/console.html ? | 19:56 |
| pleia2 | clarkb: did you want to revive this, or should I just creaet something new? https://review.openstack.org/#/c/36593/ | 19:59 |
| zul | fungi: we should probably add oslo.hacking as well | 20:00 |
| *** nayward has joined #openstack-infra | 20:00 | |
| clarkb | pleia2: restored. Feel free to push patchsets to it, I think we want a git-daemon puppet module | 20:01 |
| fungi | zul: cool--is it testing clean now? | 20:01 |
| clarkb | pleia2: then include git_daemon will install and configure it on centos | 20:01 |
| pleia2 | clarkb: thanks, will do | 20:01 |
| * pleia2 nods | 20:01 | |
| zul | fungi: almost i have 2 WIP branches | 20:01 |
| jeblair | mtreinish: we're hitting scalability limits for the way we're managing devstack nodes | 20:01 |
| fungi | zul: excellent. sounds like jog0 is very close on hacking too | 20:01 |
| clarkb | mtreinish: jeblair: that particular failure should happen a lot less often now. It is interesting that we are still seeing it | 20:02 |
| *** markmcclain has joined #openstack-infra | 20:02 | |
| jeblair | mtreinish: i'm working on addressing that; it's a multi-step process | 20:02 |
| clarkb | oh I wonder if that one failed because in progress took too long | 20:02 |
| jeblair | clarkb: that one was a build timeout on the inprogress job | 20:02 |
| jeblair | yes | 20:02 |
| clarkb | in that case different error same symptom | 20:02 |
| mtreinish | jeblair: ok, I was just curious is there a bug for recheck? or just no bug it? | 20:02 |
| clarkb | gotcha | 20:02 |
| jeblair | but same root cause | 20:02 |
| jeblair | mtreinish: may as well "no bug", we know it's a problem | 20:02 |
| reed | ttx, noticed the slowness | 20:03 |
| clarkb | I am not smart enough to grab lunch before the meeting. Going to do that now, back in a bit | 20:03 |
| reed | a lot of messages in queue | 20:03 |
| jeblair | russellb, pabelanger: if someone called in on a cell, and someone else over a landline, might the provider have used gsm for one and ulaw for the other, causing us to need to transcode? | 20:06 |
| openstackgerrit | A change was merged to openstack-infra/config: Make subunit2html.py Py3k compatible. https://review.openstack.org/39356 | 20:06 |
| anteaya | can anyone help me understand how I can add a .gitreview file to storyboard on July 16th and ttx can add a .gitreview file on July 18th, and my patch with the .gitreview file got merged but ttx is credited with the file: https://github.com/openstack-infra/storyboard/commits/master? | 20:06 |
| anteaya | did his patch overwrite mine? | 20:07 |
| openstackgerrit | A change was merged to openstack-infra/config: Create python33 gate jobs for pbr and oslo.config. https://review.openstack.org/39365 | 20:07 |
| russellb | jeblair: doubt it. asterisk is set to prefer ulaw. if the provider offers it, that's what asterisk will pick | 20:08 |
| pabelanger | or we only accept ulaw from the ITSP | 20:08 |
| ttx | anteaya: black magic | 20:08 |
| jeblair | russellb: what were we transcoding then? | 20:08 |
| russellb | and i'm betting the provider will prefer ulaw, as well (less work on their end, as it's probably received as ulaw to them) | 20:08 |
| anteaya | ttx so it appears | 20:08 |
| russellb | jeblair: ulaw <-> raw (signed linear) | 20:08 |
| russellb | jeblair: the transcoding shouldn't have been that bad | 20:08 |
| anteaya | just don't want future files disappearing, ttx, .gitreview doesn't matter but what about longer files? | 20:09 |
| jeblair | russellb: is that necessary for mixing for the conf bridge? | 20:09 |
| russellb | it was probably just the conference mixing | 20:09 |
| russellb | yes | 20:09 |
| pabelanger | russellb: I have seen issue with PLC and transcoding recently, it would be work testing that option out | 20:09 |
| russellb | have to convert it to raw signed linear before you can do mixing | 20:09 |
| pabelanger | jeblair: what are the spec's of the VM? | 20:09 |
| *** changbl has quit IRC | 20:09 | |
| ttx | anteaya: I used a chicken, a knife and a bit of your hair. | 20:09 |
| anteaya | that is where that chunk of hair went | 20:10 |
| jeblair | 2G ram, 2vcpu | 20:10 |
| jeblair | cpuinfo says model name: AMD Opteron(tm) Processor 4170 HE | 20:10 |
| russellb | pabelanger: the 2 GB one here: http://www.rackspace.com/cloud/servers/ | 20:10 |
| anteaya | ttx okay, so I'll just merge my changes you and overwrite them, okay? | 20:10 |
| pabelanger | jeblair: russellb: honestly, I think we just need to profile asterisk 11, and see how it performs. If this was 1.8, I would have a good stats on the system, but I have not rolled asterisk 11 in to production yet | 20:11 |
| ttx | anteaya: IActually I pushed a new change because the .gitreview was still not there... but I remember seeing your commit though. Weird. | 20:11 |
| pabelanger | so I am flying bling | 20:11 |
| pabelanger | blind* | 20:12 |
| jeblair | pabelanger: i imagine you flying bling now. | 20:12 |
| sdague | mordred: so this is probably as far as I can get until the pbr thing - https://review.openstack.org/#/c/39358/ | 20:13 |
| anteaya | ttx very weird | 20:13 |
| fungi | ttx: anteaya: git log --graph --oneline | 20:13 |
| fungi | ttx: anteaya: the commits were added on different branches which were later merged | 20:13 |
| russellb | pabelanger: that sounds like effort | 20:13 |
| mordred | jeblair: neat! https://review.openstack.org/#/c/37562/ | 20:13 |
| jeblair | ttx, anteaya: i'm assuming that git easily merged it because there were no conflicts. | 20:13 |
| mordred | jeblair: we added a while a gate job was running | 20:13 |
| mordred | jeblair: and the job failed to merge | 20:14 |
| pabelanger | russellb: indeed. And the major reason I have not upgraded any clients to 11 | 20:14 |
| *** changbl has joined #openstack-infra | 20:14 | |
| russellb | pabelanger: 1.8 would have been meetme, meh | 20:14 |
| anteaya | fungi: thanks | 20:14 |
| *** toddmorey has quit IRC | 20:14 | |
| russellb | pabelanger: that'd be worse | 20:14 |
| jeblair | mordred: hehe :) i love changing the rules in the middle of the game. | 20:14 |
| *** toddmorey1 has joined #openstack-infra | 20:14 | |
| anteaya | jeblair: makes sense | 20:14 |
| anteaya | so no worries about other files then since no other file would match exactly | 20:14 |
| anteaya | okay, thanks for clearing up the mystery | 20:15 |
| fungi | anteaya: virtual branches in gerrit in this case. imagine both changes were in review. one merged easily because it contained the one which was already there | 20:15 |
| pabelanger | russellb: define worse. | 20:15 |
| anteaya | ttx, all my future patches will be ringed with salt | 20:15 |
| pabelanger | have we come up with a goal of concurrent calls? | 20:15 |
| * fungi will brb | 20:15 | |
| anteaya | fungi: makes sense | 20:15 |
| jeblair | pabelanger: i think we threw 30 out as a reasonable target because that would support a board meeting | 20:16 |
| jeblair | pabelanger: (it would also support several design summit sessions, each with several participants) | 20:17 |
| pabelanger | jeblair: okay, so wouldn't be too hard to simulate that with SIPp or something and see what the load is | 20:17 |
| anteaya | ttx and do stop by once you have finished all your meetings, if you can, I have more storyboard questions | 20:17 |
| jeblair | pabelanger: i have 8 outgoing phone lines at home. :) | 20:17 |
| pabelanger | jeblair: automation FTW | 20:18 |
| jeblair | mordred: and actually if that job had existed, zuul should have run it (even if it didn't exist at the start) | 20:20 |
| mordred | jeblair: interesting | 20:22 |
| jeblair | mordred: (it didn't exist because it wasn't added to jjb, fungi just did a change that fixed that) | 20:22 |
| ttx | err, when do I sleep ? | 20:23 |
| jeblair | mordred: i'm not sure the job has been created yet though; zuul still says lost for that pbr change recheck | 20:24 |
| *** atiwari has quit IRC | 20:25 | |
| mordred | oh. right | 20:25 |
| fungi | merged at 20:07... i'll check for hung puppet on jenkins | 20:25 |
| fungi | two defunct/zombied jenkins-jobs processes as children of the puppet agent | 20:26 |
| fungi | one from the 19th, one from the 26th so days old | 20:27 |
| fungi | and otpreventing subsequent runs presumably | 20:27 |
| fungi | not preventing | 20:27 |
| jeblair | fungi: the latest batch started at 20:14 | 20:28 |
| jeblair | (of jobs) | 20:28 |
| jeblair | so it probably missed the puppet run | 20:28 |
| fungi | yeah, the config.xml is there for it now, modified at 20:25 | 20:29 |
| fungi | so should be safe to rerun | 20:30 |
| sdague | mordred: so you think update.py in requirements upstream does the oslo urls? | 20:30 |
| sdague | which change do you think fixed that? | 20:30 |
| mordred | sdague: or at least it ignores and doesn't change them? | 20:31 |
| *** dkliban is now known as dkliban_afk | 20:32 | |
| sdague | mordred: nope - http://logs.openstack.org/05/35705/8/check/gate-tempest-devstack-vm-full/37483/console.html | 20:32 |
| sdague | we'd have seen nova install 1.2.0a3 in that log if it worked | 20:33 |
| mordred | piddle | 20:33 |
| mordred | sdague: so - right now it's ignoring those wholesale | 20:33 |
| sdague | mordred: the unit tests I landed also speak otherwise | 20:35 |
| sdague | let me rebase, maybe the new stuff just has a subtle bug | 20:36 |
| *** vijendar has quit IRC | 20:37 | |
| sdague | yeh, so it ignore the url line | 20:37 |
| sdague | but it pushes the dep down | 20:38 |
| anteaya | ttx doesn't the chicken thing help with sleep? | 20:38 |
| sdague | mordred: http://paste.openstack.org/show/42670/ | 20:38 |
| *** annegentle is now known as anneisgentle | 20:40 | |
| clarkb | I am back | 20:47 |
| *** dina_belova has joined #openstack-infra | 20:48 | |
| *** dina_belova has quit IRC | 20:51 | |
| anteaya | clarkb: wb | 20:52 |
| comstud | sdague (or anyone): Can someone guide me on how to disable certain temptest tests ONLY for the cells gate? | 20:55 |
| sdague | comstud: we should talk in -qa, but we typically don't do that | 20:56 |
| jaypipes | what's the link to the reviewday for openstack? can never remember... | 20:56 |
| sdague | jaypipes: isn't off status.openstack.org now/ | 20:56 |
| comstud | sdague: ok | 20:57 |
| sdague | jaypipes: http://status.openstack.org/reviews/ | 20:57 |
| jaypipes | sdague: danke :) | 20:57 |
| *** dina_belova has joined #openstack-infra | 20:58 | |
| *** michchap has joined #openstack-infra | 20:59 | |
| *** rfolco has quit IRC | 20:59 | |
| *** nati_ueno_2 has joined #openstack-infra | 21:00 | |
| *** michchap has quit IRC | 21:00 | |
| *** michchap has joined #openstack-infra | 21:01 | |
| *** toddmorey1 has quit IRC | 21:01 | |
| *** dina_belova has quit IRC | 21:03 | |
| *** boris-42 has quit IRC | 21:05 | |
| *** bnemec has quit IRC | 21:08 | |
| *** dkranz has quit IRC | 21:10 | |
| *** bnemec has joined #openstack-infra | 21:10 | |
| anteaya | ttx when you have a lull in meetings, I just pulled down the latest storyboard changes and from what I am seeing I have to conclude that locally I can't get database columns created from the BooleanField in the story and project models | 21:11 |
| anteaya | I am looking at story.is_bug and project_milestone.released and project_milestone.undefined | 21:12 |
| anteaya | I don't know enough about django to know if this is a 1.4/1.5 issue (I am using 1.5 locally) or a db issue (I am using sqlite3) or an actual bug | 21:12 |
| *** bnemec has quit IRC | 21:13 | |
| *** bnemec has joined #openstack-infra | 21:14 | |
| clarkb | jeblair: fungi https://review.openstack.org/#/c/39137/ can we push that through? It will give us much needed wiki docs and test the update-bug script's merge hook | 21:14 |
| *** sarob_ has joined #openstack-infra | 21:15 | |
| *** bnemec has quit IRC | 21:16 | |
| jeblair | clarkb: +2 | 21:16 |
| fungi | clarkb: lgtm. merge at will | 21:16 |
| clarkb | woot thanks | 21:16 |
| fungi | i didn't approve since you may want to be in control of testing that bug update hook | 21:16 |
| clarkb | oh heh. since the new headers end in bug: #XXXXXX the existing gerrit regex works | 21:17 |
| clarkb | fungi: yup, thanks | 21:17 |
| clarkb | and by in control basically I just need to know when to check lp | 21:17 |
| ttx | anteaya: i'll have to check. I use 1.4/sqlite | 21:18 |
| ttx | anteaya: could you fire me an email so that I don't forget ? | 21:18 |
| anteaya | ttx okay, when you have a minute, thanks | 21:18 |
| anteaya | can do, where shall I send it? | 21:18 |
| *** sarob has quit IRC | 21:19 | |
| anteaya | meaning what is your email address? | 21:19 |
| openstackgerrit | A change was merged to openstack-infra/config: Add Wiki upgrade process documentation. https://review.openstack.org/39137 | 21:19 |
| anteaya | are you ttx@openstack.org? | 21:19 |
| ttx | That may work. Otherwise I'm thierry@o.o | 21:19 |
| anteaya | I'll use the thierry one | 21:20 |
| reed | looks like lists.o.o is very busy, lots of email in the queue | 21:20 |
| *** sarob_ has quit IRC | 21:20 | |
| *** russellb has quit IRC | 21:22 | |
| *** russellb has joined #openstack-infra | 21:26 | |
| *** vijendar has joined #openstack-infra | 21:27 | |
| *** vijendar has joined #openstack-infra | 21:27 | |
| jeblair | reed: oh, we should graph that with cacti | 21:31 |
| *** vipul is now known as vipul-away | 21:33 | |
| mordred | jeblair: ++ | 21:35 |
| jeblair | there are currently 73645 items in exim's queue. | 21:36 |
| clarkb | oh cacti, /me pushes patch to add pbx | 21:36 |
| fungi | mordred: https://jenkins.openstack.org/job/gate-pbr-python33/3/console "2013-07-30 21:15:26.899 | /usr/bin/python: No module named subunit" | 21:37 |
| mordred | jeblair: that's really impressive number of queue items | 21:37 |
| fungi | mordred: no subunit in python 3.3? | 21:37 |
| *** SergeyLukjanov has quit IRC | 21:37 | |
| mordred | fungi: h... | 21:37 |
| clarkb | fungi: subunit should python3.3 | 21:37 |
| mordred | 2013-07-30 21:15:26.091 | python-subunit==0.0.13 | 21:37 |
| *** mriedem has quit IRC | 21:37 | |
| fungi | yeah, so... wth? | 21:38 |
| jeblair | mordred: and some of them aren't going to gmail | 21:38 |
| openstackgerrit | Clark Boylan proposed a change to openstack-infra/config: Add pbx and new logstash workers to cacti. https://review.openstack.org/39374 | 21:38 |
| fungi | oh! mordred python-subunit installed for python3.3 but this is being run with /usr/bin/python instead | 21:38 |
| clarkb | jeblair: I want to say I haven't seen horrible delays from the general list with my gmail account | 21:39 |
| mordred | fungi: ah! so we're missing subunit from the 33 slave hosts | 21:40 |
| * fungi checks | 21:41 | |
| fungi | mordred: looks like we don't pip install it on slaves (slave.pp has ensure absent for it) | 21:42 |
| jeblair | reed: we're running a real mailing list server now! exim needs tuning! | 21:42 |
| reed | jeblair, I figured you knew the answer :) | 21:42 |
| clarkb | fungi: mordred: ya we rely on the test venv to provide the deps | 21:42 |
| jeblair | reed: there are two things we should do, which are relatively easy... | 21:43 |
| clarkb | fungi: mordred: I think the bug here is that the run-tox script is naieve and assumes /usr/bin/python is THE python | 21:43 |
| mordred | ah. | 21:43 |
| fungi | clarkb: yep, fixing | 21:43 |
| jeblair | reed: first, run the queue processor more often (like 5 minutes instead of 30) | 21:43 |
| jeblair | reed: second, think about how many messages exim should accept from mailman at once | 21:44 |
| *** vipul-away is now known as vipul | 21:44 | |
| jeblair | reed: i'll try to do at least the first one real soon, and start thinking about the second | 21:44 |
| reed | jeblair, wonderful | 21:44 |
| *** pabelanger has quit IRC | 21:45 | |
| jeblair | so my desktop just crashed (cosmic ray and no ecc ram, i assume) | 21:46 |
| jeblair | which is no big deal, but it has given men an opportunity to experience a new behavior from git! | 21:46 |
| jeblair | fatal: loose object 03201aa88167a070d44b3850609ecbb1041f9d9c (stored in .git/objects/03/201aa88167a070d44b3850609ecbb1041f9d9c) is corrupt | 21:46 |
| mordred | jeblair: nice! | 21:47 |
| jeblair | apparently recovering from that is hard. | 21:47 |
| jeblair | http://git.kernel.org/cgit/git/git.git/tree/Documentation/howto/recover-corrupted-blob-object.txt?id=HEAD | 21:47 |
| jeblair | possible, but hard. | 21:47 |
| clarkb | :/ | 21:47 |
| mordred | jeblair: is the solution re-clone? | 21:48 |
| jeblair | that is what i am going to do. all my work is in gerrit anyway. | 21:48 |
| fungi | wow, a "loose object" sounds fun | 21:48 |
| jeblair | fungi: i think that just means it's not in a packfile | 21:49 |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/config: Validate that zuul-referenced jobs exist https://review.openstack.org/38755 | 21:50 |
| fungi | clarkb: so on deeper inspection, i'm not seeing where it's calling the system default python... run-tox.sh invokes .tox/$venv/bin/python when calling testr list-tests | 21:51 |
| fungi | so could testr list-tests have a /usr/bin/python assumption embedded somewhere? | 21:51 |
| clarkb | maybe? I think entry points are used to make the magic happen | 21:52 |
| *** nati_ueno_2 has quit IRC | 21:52 | |
| clarkb | its possible that the pip installing the things is running under a different python than python3.3? | 21:52 |
| clarkb | fungi: yeah loose files and pack files are the two types of storage git uses | 21:53 |
| clarkb | loose is unpacked :) | 21:53 |
| fungi | aha | 21:53 |
| fungi | clarkb: the pip installing the things (according to freeze) is .tox/py33/bin/pip | 21:54 |
| fungi | which does indicate python-subunit is in there | 21:54 |
| fungi | so i think the testr subcommand may be the one making a bad assumption here | 21:54 |
| fungi | i'll see if i can track it down | 21:54 |
| fungi | after dinner | 21:55 |
| *** bnemec has joined #openstack-infra | 21:56 | |
| *** bnemec has quit IRC | 21:58 | |
| *** lcestari has quit IRC | 21:58 | |
| *** dina_belova has joined #openstack-infra | 21:59 | |
| *** bnemec has joined #openstack-infra | 21:59 | |
| nati_ueno | hi! infra team. May I ask your helps? I'm working on two critical bug which fails neutron-gating. but I can't reproduce it on my local. so I wanna use one VM on the gating env.. | 21:59 |
| *** bnemec has quit IRC | 22:01 | |
| fungi | nati_ueno: have you seen the readme in https://github.com/openstack-infra/devstack-gate/ ? there's a section called "Simulating Devstack Gate Tests" which provides a walkthrough | 22:02 |
| *** burt has quit IRC | 22:02 | |
| *** bnemec has joined #openstack-infra | 22:02 | |
| nati_ueno | fungi: I'm using same localrc | 22:03 |
| fungi | nati_ueno: 8gb ram? | 22:03 |
| *** dina_belova has quit IRC | 22:03 | |
| nati_ueno | fungi: ah sorry no | 22:03 |
| clarkb | I think you need at least 4gb | 22:04 |
| fungi | ubuntu precise 12.04 lts, plus some puppeting which is described in that writeup | 22:04 |
| *** bnemec has quit IRC | 22:04 | |
| fungi | that readme also shows you how to find which environment variables you want to set to configure the tests themselves, and then launching them from devstack-vm-gate-wrap.sh like the gate jobs do | 22:05 |
| nati_ueno | ok let's me try | 22:06 |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/config: Add queue runner interval option to exim https://review.openstack.org/39377 | 22:07 |
| fungi | nati_ueno: on the vmt we use those same steps to test embargoed security fixes since we can't test them in public prior to announcing, so that's as close to the automated jobs as we can get it | 22:07 |
| jeblair | reed: item 1: https://review.openstack.org/39377 | 22:07 |
| nati_ueno | fungi: sure | 22:08 |
| *** changbl has quit IRC | 22:08 | |
| *** toddmorey has joined #openstack-infra | 22:08 | |
| reed | cool | 22:09 |
| *** vipul is now known as vipul-away | 22:11 | |
| *** vipul-away is now known as vipul | 22:12 | |
| clarkb | jeblair: fungi mordred https://review.openstack.org/#/c/39379/ | 22:13 |
| clarkb | wow I derped hard submitting that | 22:16 |
| sdague | ttx: you still awake? | 22:17 |
| clarkb | but all is better now | 22:17 |
| *** _TheDodd_ has quit IRC | 22:18 | |
| clarkb | jeblair: https://review.openstack.org/#/c/39377/ has tyops I think | 22:22 |
| clarkb | I left a review | 22:22 |
| fungi | clarkb: on 39379 should we merge the abstract into the readme instead? | 22:22 |
| clarkb | fungi: we can do that if you think it will be easier to find | 22:22 |
| fungi | clarkb: well, the intro paragraph for the readme was intended to be reusable as a short abstract, so i wonder if we can have something variable-length in there to be able to meet submission requirements for various fora | 22:23 |
| clarkb | oh I didn't realize that the readme wsa intended to eb used that way | 22:24 |
| *** michchap has quit IRC | 22:24 | |
| clarkb | that'll teach me | 22:24 |
| * fungi is thinking forward to starting to curate these and how we can standardize the repos for our presentations down the road | 22:24 | |
| *** toddmorey has quit IRC | 22:24 | |
| *** michchap has joined #openstack-infra | 22:24 | |
| clarkb | ya, its a good idea | 22:24 |
| clarkb | fungi: perhaps an Abstract section in the rst so that it is unambiguous? and I don't think we will end up with something that is one size fits all, but should have something that sets the tone in the general case | 22:26 |
| fungi | that makes great sense | 22:26 |
| *** toddmorey has joined #openstack-infra | 22:26 | |
| *** markmcclain has quit IRC | 22:26 | |
| fungi | i'm still struggling with the publish job for the repo though... it seems to keep dumping content into $ZUUL_BRANCH instead of the actual branch name | 22:27 |
| fungi | i'm clearly missing something to get that passed in | 22:27 |
| clarkb | is $ZUUL_BRANCH being used literally and not interpolated? | 22:27 |
| fungi | it's literally going into a directory on the docs site called /web/content/infra/publications/$ZUUL_BRANCH/ | 22:28 |
| *** jhesketh has joined #openstack-infra | 22:28 | |
| *** avtar has quit IRC | 22:29 | |
| clarkb | fungi: have a link handy to one of the job runs? | 22:29 |
| mgagne | under which user is jenkins-job-update running on the master? my "question" is regarding the fact that jenkins::job_builder requires the jenkins user to exist on the node hosting/running JJB which might not always be the case from a reusability pov. | 22:29 |
| * fungi will find | 22:29 | |
| fungi | clarkb: https://jenkins.openstack.org/job/infra-publications-publish/10/console | 22:31 |
| *** rnirmal has quit IRC | 22:32 | |
| *** dkliban_afk has quit IRC | 22:32 | |
| clarkb | fungi: https://jenkins.openstack.org/job/infra-publications-publish/10/parameters/? I think ZUUL_BRANCH may not be provided in the post queue anymore | 22:33 |
| clarkb | we should probably fix that | 22:33 |
| jeblair | clarkb, fungi: use ZUUL_REF instead | 22:33 |
| fungi | oh! | 22:34 |
| clarkb | aha | 22:34 |
| fungi | jeblair: thanks!!! | 22:34 |
| fungi | mgagne: root apparently... https://github.com/openstack-infra/config/blob/master/modules/jenkins/manifests/job_builder.pp#L42 | 22:34 |
| fungi | mgagne: we don't seem to specify a user to run that exec under | 22:34 |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/zuul: Support multiple triggers https://review.openstack.org/39186 | 22:35 |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/zuul: Add a timer trigger https://review.openstack.org/39187 | 22:35 |
| jeblair | clarkb, fungi, mordred: ^ both are ready for review now | 22:35 |
| clarkb | jeblair: cool will put that at the top of the list | 22:36 |
| jeblair | clarkb, fungi, mordred: note there's a backwards incompat layout.yaml change needed | 22:36 |
| clarkb | fungi: I updated the publications change. Not sure how we want to merge the content I wrote if at all, so it is there along with the old content for reviewers | 22:36 |
| clarkb | fungi: I constructed the stuff I wrote based on some old abstracts jeblair and mordred used then made them more modern | 22:36 |
| fungi | clarkb: heh... i guess i should have looked at https://jenkins.openstack.org/job/infra-publications-publish/10/parameters/ | 22:36 |
| mgagne | fungi: my "issue" is that I can't include (reuse) jenkins::job_builder without the jenkins user which is not reasonable IMO. So I'm trying to see if I could propose removing the dependency on the jenkins user in jenkins::job_builder. | 22:37 |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/config: Add queue runner interval option to exim https://review.openstack.org/39377 | 22:38 |
| clarkb | mgagne: yes, I think making the user configurable is a good thing | 22:38 |
| jeblair | http://logs.openstack.org/55/38755/3/check/gate-config-layout/2641/console.html | 22:38 |
| jeblair | clarkb, fungi, mordred: ^ you should all look at the bottom of that. :) | 22:38 |
| *** nayward has quit IRC | 22:38 | |
| *** vipul is now known as vipul-away | 22:39 | |
| fungi | mgagne: clarkb: agreed, the username does not need hardcoding | 22:39 |
| clarkb | jeblair: that is really cool | 22:39 |
| fungi | poor gerritbot, being made an example of | 22:40 |
| clarkb | jeblair: https://review.openstack.org/#/c/39379/2 in case you missed it (publications isn't spamming in here) | 22:40 |
| mgagne | fungi: Can I propose removing it altogether or is there a technical/security reason for it to be present? | 22:40 |
| clarkb | mgagne: you need a user to create jobs | 22:40 |
| clarkb | mgagne: it is authenticating with jenkins | 22:40 |
| mgagne | clarkb: this user isn't the API user, it's the owner of the JJB config file | 22:40 |
| jeblair | clarkb: i think mgagne is referring to the os user | 22:40 |
| mgagne | jeblair: yes | 22:40 |
| clarkb | oh we are talking about puppet here | 22:41 |
| jeblair | yep | 22:41 |
| mgagne | clarkb: this line: https://github.com/openstack-infra/config/blob/master/modules/jenkins/manifests/job_builder.pp#L57 | 22:41 |
| fungi | mgagne: so, that file contains authentication data | 22:41 |
| fungi | mgagne: we need to set it owned by the account which will need to read the file | 22:42 |
| fungi | mgagne: so that we can restrict permissions on it appropriately | 22:42 |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/config: Validate that zuul-referenced jobs exist https://review.openstack.org/38755 | 22:42 |
| jeblair | fungi: but that account is root, right? | 22:42 |
| *** prad_ has quit IRC | 22:42 | |
| fungi | jeblair: ah! right, i went in a circle there | 22:42 |
| mgagne | fungi: it is true. I was wondering if the said user is actually jenkins or root. | 22:43 |
| fungi | mgagne: root, okay so yes we can change that without causing any issues i believe | 22:43 |
| * fungi is scattered | 22:43 | |
| *** vipul-away is now known as vipul | 22:43 | |
| fungi | mgagne: we presumably want to set owner => 'root', in that case | 22:44 |
| *** rcleere has quit IRC | 22:44 | |
| *** jpich has quit IRC | 22:44 | |
| *** toddmorey has quit IRC | 22:44 | |
| jeblair | fungi: or omit it? | 22:44 |
| fungi | since puppet will default to using the ownership of the files as they exist on the master's filesystem in lieu of any specified owner | 22:44 |
| * fungi discovered this the fun way a couple months ago | 22:45 | |
| jeblair | ah | 22:45 |
| *** anteaya has quit IRC | 22:45 | |
| mgagne | fungi: this would break backward compatibility assuming that outsiders were relying on this particular ownership to be set. So I'm wondering if it's actually a good idea to propose such change. :-/ | 22:45 |
| fungi | mgagne: this would only have worked for them under two possible situations... running puppet as root or running puppet as a user named "jenkins" since the file isn't readable by anyone else | 22:46 |
| mgagne | fungi: or to some extend, setup a cron job under the jenkins user which might be unlikely. | 22:47 |
| fungi | it seems a little unlikely that they had their puppetry applied by the jenkins user (it would have required non-trivial modifications elsewhere on the system to be able to write to the correct places) | 22:47 |
| *** toddmorey has joined #openstack-infra | 22:47 | |
| fungi | yes, or they did other stuff not puppeted there | 22:47 |
| *** EmilienM has quit IRC | 22:47 | |
| fungi | either way, it seems pretty unlikely to impact anyone negatively | 22:48 |
| sdague | mordred: so there is an issue on requirements... | 22:48 |
| sdague | in that I know it's wrong, which is part of the problem. Like the fact that nova can't function with sqla 0.7 | 22:49 |
| clarkb | jeblair: thank you for the review, updated to address your comments and now I am going to deep dive into zuul review | 22:51 |
| openstackgerrit | Jeremy Stanley proposed a change to openstack-infra/config: Publications use ZUUL_REF instead of ZUUL_BRANCH. https://review.openstack.org/39382 | 22:51 |
| *** EmilienM has joined #openstack-infra | 22:52 | |
| jeblair | clarkb: i'm just trying to learn how to include asterisk in sphinx docs. | 22:53 |
| jeblair | then the timer change should be good | 22:53 |
| clarkb | is that what caused the -1? | 22:53 |
| jeblair | yeah | 22:53 |
| openstackgerrit | Mathieu Gagné proposed a change to openstack-infra/config: Set ownership of JJB config file to root https://review.openstack.org/39383 | 22:55 |
| *** toddmorey has quit IRC | 22:57 | |
| *** krtaylor has quit IRC | 22:57 | |
| *** dina_belova has joined #openstack-infra | 22:59 | |
| *** vipul is now known as vipul-away | 23:01 | |
| fungi | backslashes are supposed to be able to escape asterisks in rest/sphinx | 23:01 |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/zuul: Add a timer trigger https://review.openstack.org/39187 | 23:01 |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/config: Validate that zuul-referenced jobs exist https://review.openstack.org/38755 | 23:02 |
| *** dina_belova has quit IRC | 23:03 | |
| openstackgerrit | A change was merged to openstack-dev/pbr: Remove missing_reuqires optimization https://review.openstack.org/39330 | 23:04 |
| *** emagana has quit IRC | 23:05 | |
| *** nati_ueno_2 has joined #openstack-infra | 23:12 | |
| *** sandywalsh has quit IRC | 23:14 | |
| *** marun has joined #openstack-infra | 23:16 | |
| marun | mordred: ping | 23:16 |
| mordred | sdague: oh! | 23:20 |
| mordred | sdague: well, that's a good discover! | 23:20 |
| mordred | y | 23:20 |
| sdague | yeh, there's a lot of things like that | 23:20 |
| sdague | our minimums are "interesting" | 23:21 |
| sdague | and largely overriden by projects | 23:21 |
| clarkb | jeblair: I have reviewed both changes | 23:22 |
| sdague | anyway, our actual impediment right now is oslo pre-releases | 23:22 |
| sdague | I'm trying to figure out if we just try to get with markmc in the morning, or drive this to the list | 23:23 |
| *** reed has quit IRC | 23:23 | |
| mordred | what's up with oslo pre-relase? | 23:23 |
| sdague | so nova needs oslo 1.2.0a3 | 23:23 |
| sdague | will not work without it | 23:23 |
| sdague | will not start without it | 23:24 |
| *** reed has joined #openstack-infra | 23:24 | |
| sdague | so, if it was actually a released thing, we could just bump global requirements | 23:24 |
| sdague | but it's not | 23:24 |
| mordred | but we've got this middle ground thing | 23:24 |
| mordred | yeah | 23:24 |
| sdague | right | 23:24 |
| mordred | so - amongst the things I want to do: | 23:24 |
| sdague | so to me it feels like we need to either use releases | 23:24 |
| mordred | a) add oslo.config master to the gate | 23:25 |
| sdague | or use the oslo.config git tree | 23:25 |
| mordred | so - honestly, the pattern we have with every other library is working | 23:26 |
| mordred | gate on the git version in devstack | 23:26 |
| mordred | and put the released version in requirements without an upper bound | 23:26 |
| sdague | right | 23:26 |
| sdague | ok, so we could do that | 23:26 |
| mordred | there is a theoretical straw man that we might want to break something | 23:26 |
| *** mrodden has quit IRC | 23:27 | |
| mordred | except, I think we've past the point where that's ok | 23:27 |
| mordred | we are too complex to allow that | 23:27 |
| sdague | yeh | 23:27 |
| mordred | and too many people consume our trunk | 23:27 |
| mordred | including us | 23:27 |
| sdague | the neutroclient thing is going to get interesting though | 23:27 |
| sdague | they want to break compat with 3 | 23:27 |
| clarkb | uh | 23:28 |
| clarkb | I thought we don't do that with our clients | 23:28 |
| mordred | sdague: no. they cannot do that | 23:28 |
| mordred | it's just - we're too big and complex | 23:28 |
| sdague | um... I got push back in removing that cap | 23:29 |
| sdague | and I caved | 23:29 |
| sdague | so if you want to bring it up, go for it | 23:29 |
| mordred | heh | 23:29 |
| mordred | so, what I mean is - | 23:29 |
| sdague | but the neutron cap is still in there | 23:29 |
| sdague | anyway, if we are good at putting oslo master into devstack, we can solve this all tomorrow | 23:30 |
| sdague | oslo.config master | 23:30 |
| mordred | I think we should absolutely put oslo.config master in devstack | 23:30 |
| mordred | and oslo.messaging | 23:30 |
| *** UtahDave has quit IRC | 23:30 | |
| sdague | ok, lets do it. That's a plan | 23:30 |
| mordred | awesome. | 23:30 |
| mordred | I feel like I wrote that patch once already | 23:30 |
| sdague | we need markmc to weigh in on this? | 23:30 |
| sdague | or is it just fiat, because you either get to have releases, or we have your git tree | 23:31 |
| mordred | https://review.openstack.org/#/c/37295/ | 23:31 |
| mordred | I think we can do both-ish | 23:31 |
| mordred | devstack _has_ to consume trunk of all of our related things | 23:31 |
| mordred | otherwise things break | 23:31 |
| mordred | however, if you want nova to consume tarball prereleases for unittests - fine | 23:32 |
| mordred | because we should have already proven that that code is fine for consumption by everyone else in the devstack gate | 23:32 |
| mordred | right? | 23:32 |
| clarkb | jeblair: I think that one more elasticsearch node may be necessary in the near future, but not due to panic and attempting to scale. http://cacti.openstack.org/cacti/graph.php?action=view&local_graph_id=686&rra_id=all indexes are huge and we are keeping 2 weeks of them | 23:32 |
| sdague | yeh... still ceilo is weird :) | 23:32 |
| mordred | sdague: what's up with ceilo? | 23:32 |
| mordred | do we need to rally dhellman and jd__ ? | 23:32 |
| sdague | consuming nova tarball in unit tests | 23:32 |
| clarkb | jeblair: in related news elasticsearch did not die last week and mostly kept up with demand. I think we finally have a thing that is looking stable | 23:32 |
| sdague | no, not weird as a problem | 23:32 |
| sdague | oh... and one last thing | 23:33 |
| *** nati_ueno_2 has quit IRC | 23:33 | |
| sdague | unrelated | 23:33 |
| sdague | catching stack traces in logs | 23:33 |
| mordred | yes to both | 23:33 |
| mordred | well, also | 23:33 |
| mordred | glance | 23:33 |
| sdague | so recently the ceilo patches that went into devstack basically just meant 100 new ceilo stack traces in the logs | 23:34 |
| mordred | glance does a bunch of functional/integration stuff in their unittest suite | 23:34 |
| mordred | I'd like to move that to be more like the swift model | 23:34 |
| clarkb | oh I need to add ceilometer logs at some point | 23:34 |
| clarkb | probably after the disk situation is stable | 23:34 |
| mordred | did you see that we're running swift functional tests from their tree against devstack installs now? | 23:34 |
| sdague | clarkb: well, this was actually nova logs | 23:34 |
| sdague | mordred: yep | 23:34 |
| clarkb | sdague: oh neat | 23:34 |
| sdague | because ceilo installs itself into nova | 23:34 |
| clarkb | mordred: and we need to hunt down chmouel and notmyname so that we can make them voting and run them in the gate | 23:34 |
| mordred | sdague: I want to do that with ceilometer and glance in their need to consume things - and keystone with their multi-version stuff | 23:35 |
| sdague | right, that's cool | 23:35 |
| mordred | clarkb: k. well, the swift team seems pleased with them | 23:35 |
| *** pcrews has quit IRC | 23:35 | |
| mordred | in fact, asked if we could add their probe tests | 23:35 |
| mordred | :) | 23:35 |
| sdague | but something I want to figure out if we can get post H3 is at least ensuring we get no worse on stack tracing | 23:35 |
| *** vipul-away is now known as vipul | 23:35 | |
| clarkb | mordred: maybe we will have better luck if I write teh change and get notmyname to +1 | 23:36 |
| * clarkb does that | 23:36 | |
| dhellmann | sdague: please open a bug against ceilometer with some details? | 23:36 |
| sdague | there is no way we're going to get those all out of there, but we could make a white list of what's there | 23:36 |
| mordred | sdague: +100 | 23:36 |
| sdague | and not let in anything new | 23:36 |
| mordred | dhellmann: we were just talking about ways to get ceilometer to not need to consume nova tarball in its tree | 23:36 |
| dhellmann | I would love to hear the answer to that | 23:37 |
| mordred | oh god | 23:37 |
| marun | mordred: unrelated, when you have a moment. | 23:37 |
| mordred | someone just filed a bug about versioning | 23:37 |
| openstackgerrit | James E. Blair proposed a change to openstack-infra/zuul: Make a test more robust https://review.openstack.org/39387 | 23:37 |
| mordred | "glance version is not monotonic" | 23:37 |
| dhellmann | mordred: we need nova to test the notifier plugin we have, but that's a tiny number of tests | 23:37 |
| sdague | dhellmann: http://logs-dev.openstack.org/21/37821/15/check/gate-tempest-devstack-vm-full/37529/logs/screen-n-cpu.txt.gz#_2013-07-30_22_31_29_407 | 23:37 |
| * mordred goes to mark notabug | 23:37 | |
| openstackgerrit | Clark Boylan proposed a change to openstack-infra/config: Vote and gate on swift functional tests. https://review.openstack.org/39388 | 23:37 |
| mordred | marun: what's up? | 23:37 |
| dhellmann | sdague: that's choking my browser :-/ | 23:38 |
| sdague | dhellmann: works in chrome on linux :) | 23:38 |
| marun | mordred: I'm seeing an issue with pbr and neutron. Since the gate is presumably passing it may be rhel specfic. | 23:38 |
| sdague | it takes a little while | 23:38 |
| sdague | it's 35 MB of html | 23:38 |
| dhellmann | sdague: ouch | 23:38 |
| * dhellmann downloads instead | 23:38 | |
| clarkb | dhellmann: I have had good luck opening those files with vim. I imagine emacs does a decent job if you prefer that editor | 23:38 |
| marun | mordred: https://github.com/openstack-dev/pbr/commit/8e58c2fa58fd1aa6f9985dcb4e210508a73e1df7#L1R455 | 23:38 |
| marun | mordred: https://github.com/openstack/neutron/blob/master/setup.cfg#L76 | 23:39 |
| *** mrodden has joined #openstack-infra | 23:39 | |
| sdague | yeh, also if you wget it comes as txt because of content negotiation | 23:39 |
| marun | mordred: the second link is to a line for neutron's setup.cfg that includes a class and method | 23:39 |
| mordred | marun: hrm.... | 23:40 |
| clarkb | fungi: nice catch on the zuul layout sample that was missing the gerrit trigger | 23:40 |
| marun | mordred: the first link is to the line that is only putting the class name into the generated script | 23:40 |
| dhellmann | sdague: got it | 23:40 |
| dhellmann | sdague: nova is reporting an instance that libvirt does not know about | 23:40 |
| dhellmann | sdague: or one without a vm id? | 23:40 |
| marun | mordred: so the result is sys.exit(Dnsmasq()) instead of sys.exit(Dnsmasq.lease_update()) | 23:40 |
| dhellmann | something like that | 23:41 |
| *** dims has quit IRC | 23:41 | |
| mordred | marun: yup. so - a) nice catch b) I'm assuming that Dnsmasq.lease_update() will work ... | 23:41 |
| sdague | dhellmann: you shouldn't be filling up n-cpu with stack traces on that | 23:41 |
| dhellmann | indeed | 23:41 |
| dhellmann | why does nova want to send a notification for a vm that doesn't exist? | 23:41 |
| mordred | marun: we should change this now that we're not actually using entry_points .. I have NO IDEA what changing that line will do on windows though | 23:41 |
| marun | mordred: I don't know how this managed to pass the gate | 23:42 |
| dhellmann | sdague: I'll open a bug | 23:42 |
| mordred | marun: do we test the executable? | 23:42 |
| marun | mordred: Or are the recent quantum failures masking this problem maybe? | 23:42 |
| mordred | marun: oh - yeah. we don't gate on quantum currently | 23:42 |
| mordred | totally masking the problem | 23:42 |
| marun | mordred: ah, fsck, that explains it. | 23:42 |
| mordred | sorry neutron | 23:42 |
| clarkb | I am going to start referring to it as Jimmy | 23:43 |
| clarkb | and pretend neutron is a surname | 23:43 |
| marun | mordred: yeah, the dhcp agent doesn't work due to the script generating an exception, so vm's won't come up properly. | 23:43 |
| dhellmann | sdague: are these tests running on a system with a real libvirt, or a fake virt driver? | 23:43 |
| sdague | dhellmann: this is real libvirt | 23:43 |
| clarkb | dhellmann: should be real libvirt just no kvm | 23:43 |
| sdague | this is a successful tempest run | 23:43 |
| sdague | so over the course of that we built 80ish libvirt guests | 23:44 |
| sdague | for reals | 23:44 |
| dhellmann | https://bugs.launchpad.net/ceilometer/+bug/1206731 | 23:44 |
| uvirtbot | Launchpad bug 1206731 in ceilometer "notifier should not log traceback if VM does not exist" [Undecided,New] | 23:44 |
| marun | mordred: do you want to submit a fix for this? either take the raw value that appears in setup.cfg or concatenate ep.attrs with '.' | 23:44 |
| marun | clarkb: +1 | 23:45 |
| marun | (to jimmy) | 23:45 |
| clarkb | :) | 23:45 |
| *** sandywalsh has joined #openstack-infra | 23:46 | |
| sdague | dhellmann: I think the vm exists | 23:47 |
| sdague | I think you are using the wrong parameter to look it up | 23:47 |
| dstufft | packaging would be atleast 10x easier if we fired windows off into the sun | 23:47 |
| dhellmann | sdague: yeah, that may be -- I added a comment to the ticket to that effect | 23:47 |
| dhellmann | sdague: let me rephrase the description | 23:47 |
| sdague | mordred: ok, back to stack trace white listing | 23:48 |
| dhellmann | sdague: smells like dinner is ready, I'll make sure jd__ sees that ticket | 23:48 |
| sdague | cool | 23:48 |
| sdague | where in the pipeline should we put this? | 23:48 |
| sdague | because I'm going after this as soon as I get the requirements stuff nailed | 23:49 |
| *** dhellmann is now known as dhellmann_ | 23:50 | |
| jeblair | sdague: have d-g run it after the end of the tests, and the fail if there are new ones? | 23:50 |
| sdague | jeblair: could, I might ask for +2 on d-g if we put it there | 23:50 |
| clarkb | throwing this out there, what about running a periodic job that did batch type processing on elasticsearch? | 23:51 |
| jeblair | clarkb: i think sdague wants to gate | 23:51 |
| sdague | clarkb: we need to gate on it | 23:51 |
| clarkb | I see | 23:51 |
| jeblair | clarkb: there's an open openstack-ci bug about it, even. :) | 23:51 |
| clarkb | and that is probably a good idea | 23:51 |
| clarkb | because otherwise no one will respect it | 23:51 |
| jeblair | sdague: do you expect the whitelist to change often? | 23:52 |
| sdague | because..... the bug dhellmann just filed, as an example | 23:52 |
| mordred | dstufft: I do not know why people think that we should expect packaging to work seamless and the same way on windows | 23:52 |
| clarkb | ya, definitely gate on it | 23:52 |
| sdague | jeblair: well, I'm hoping it goes away over time | 23:52 |
| sdague | but I can imagine that a flakey scenario where we need to let one in | 23:52 |
| sdague | until we can figure it out | 23:52 |
| dstufft | mordred: I do not know why people expect packaging to work | 23:53 |
| dstufft | ;) | 23:53 |
| bodepd | I noticed that the zuul module uses the template param from apache::vhost which has sense beebn deprecated | 23:54 |
| mordred | bodepd: I believe that all of our use of the apache module is VERY old | 23:54 |
| bodepd | did you have specific plans for how you were going to refactor when you want to update? | 23:54 |
| clarkb | bodepd: ya, we are running the first tagged version of the apache module. So much stuff has changed in incompatible ways | 23:55 |
| mordred | nope | 23:55 |
| bodepd | cool. just asking. | 23:55 |
| mordred | we were just talking yesterday about how we'd like for our different apache usage to be more uniform and not-bat-shit | 23:55 |
| mgagne | bodepd: I think that infra is using an old version of puppetlabs-apache | 23:55 |
| mgagne | bodepd: 0.0.4 to be more specific =) | 23:55 |
| bodepd | not surprised, just seeing if you guys had a fix in mind before I start fixing things | 23:56 |
| sdague | jeblair: ok, this will need to be python code, we'll need some reasonably complex matching | 23:56 |
| clarkb | bodepd: we don't... | 23:56 |
| bodepd | trying to update zuul so that I will start using gearman | 23:56 |
| sdague | you ok with that in d-g? | 23:56 |
| clarkb | however if you have suggestions I am very open to them | 23:57 |
| marun | mordred: I'd like to make sure the pbr issue gets covered. Do you want to take responsibility for getting a fix in or should I do it? | 23:57 |
| clarkb | unfortunately we can't upgrade piecemeal unless we do something nasty like install a newer version under a different name | 23:57 |
| sdague | here is our current TRACE count | 23:57 |
| sdague | http://paste.openstack.org/show/42690/ | 23:57 |
| sdague | 127 TRACES found in screen-n-cpu.txt.gz | 23:57 |
| jeblair | sdague: yeah, i think that's the right place for it. | 23:57 |
| mordred | marun: if you file a bug, I will get to it eventually, but it's a crazy day. if it's something like you feel like you could fix in a few minutes, I will definitely fast-track your patch | 23:57 |
| bodepd | dealing with the upstream backwards compat changes is difficult and taken up way too much of my time lately | 23:57 |
| marun | mordred: ok, I'll take it on. | 23:58 |
| bodepd | my first guess would be just to use a regular file resource. I'll have a poke | 23:58 |
| sdague | actually.... | 23:58 |
| sdague | jeblair: would you mind if we put the tool in tempest, and called it from d-g? | 23:58 |
| jeblair | sdague: that works too. | 23:58 |
| sdague | because there was desire to have this in the stress tests (basically stress test passes if no stack traces) | 23:59 |
| sdague | ok, I'll go down that path | 23:59 |
| jeblair | sdague: that actually sounds better now that i think about it. | 23:59 |
| mordred | bodepd: honestly, if you have good thoughts on it - I'm _sure_ we'll think whatever your idea is is the right one :) | 23:59 |
| mordred | jeblair, sdague ++ | 23:59 |
| mordred | marun: thank you!!! | 23:59 |
| sdague | ok, it's a plan | 23:59 |
| marun | mordred: np | 23:59 |
| jeblair | (make it easily available for anyone who runs tests in any context) | 23:59 |
| *** dina_belova has joined #openstack-infra | 23:59 | |
| sdague | jeblair: yep | 23:59 |
| mordred | marun: I'm wanting to make another pbr release soon anyway - so we should be able to get it out the door quickly | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
