*** sanchitmalhotra1 has joined #openstack-swift | 00:31 | |
kota_ | good morning | 00:31 |
---|---|---|
*** sanchitmalhotra has quit IRC | 00:31 | |
*** sanchitmalhotra1 is now known as sanchitmalhotra | 00:31 | |
mattoliverau | kota_: morning | 00:55 |
*** gyee has quit IRC | 00:59 | |
kota_ | mattolivearu | 01:06 |
kota_ | :o/ | 01:06 |
kota_ | unexpected pushing to enter key :/ | 01:06 |
*** sanchitmalhotra1 has joined #openstack-swift | 01:31 | |
*** sanchitmalhotra has quit IRC | 01:31 | |
*** sanchitmalhotra1 is now known as sanchitmalhotra | 01:31 | |
*** dfflanders has joined #openstack-swift | 01:40 | |
zhengyin | kota_,mattoliverau:morning | 01:40 |
mattoliverau | zhengyin: o/ | 01:41 |
*** rickyrem has joined #openstack-swift | 02:04 | |
*** rickyrem has quit IRC | 02:06 | |
openstackgerrit | zheng yin proposed openstack/swift: add assert error information https://review.openstack.org/379428 | 02:19 |
zhengyin | mattoliverau: hi, mattoliverau, are you here ? | 02:41 |
mattoliverau | zhengyin: yes.. And no, just at lunch. What's up? | 02:42 |
zhengyin | mattoliverau: Do you have a time to help me review a patch 355858? :) | 02:43 |
patchbot | https://review.openstack.org/#/c/355858/ - swift - simplify code | 02:43 |
*** abhitechie has joined #openstack-swift | 02:44 | |
*** dmorita has quit IRC | 02:54 | |
*** david-lyle has quit IRC | 03:04 | |
*** tongli_ has joined #openstack-swift | 03:08 | |
*** dmorita has joined #openstack-swift | 03:09 | |
*** tongli has quit IRC | 03:10 | |
openstackgerrit | zheng yin proposed openstack/swift: fix word spelling mistake https://review.openstack.org/379909 | 03:14 |
*** dmorita has quit IRC | 03:14 | |
*** aswadr_ has joined #openstack-swift | 03:24 | |
notmyname | hello, world | 03:31 |
*** abhitechie has quit IRC | 03:38 | |
*** links has joined #openstack-swift | 03:58 | |
*** lhx_ has joined #openstack-swift | 04:08 | |
jrichli | notmyname: I am reading over all the TC candidacy mails, and re-reading the TC meetings involving the golang decision. | 04:11 |
jrichli | Whew! It's a lot to go through. but I am going to make a very informed decision when making my 6 selections! | 04:12 |
notmyname | :-) | 04:14 |
zaitcev | meaning revenge? | 04:15 |
jrichli | no, but its a good discussion to look over to get a feel for how people think and what they value in openstack | 04:15 |
notmyname | there's a lot of people running. 14 or 15 | 04:16 |
*** mingyu has joined #openstack-swift | 04:17 | |
jrichli | interesting. maybe a sign that people want to get more involved | 04:17 |
notmyname | I think it's a sign of rising interest in what the TC is doing (and perhaps some discomfort with recent decisions the TC has made) | 04:18 |
jrichli | well stated. i wasn't sure how to say that | 04:19 |
jrichli | I gotta met this cdent at the summit. I feel like giving him a high five or something | 04:21 |
openstackgerrit | Merged openstack/swift: Add more comment to authtoken sample options https://review.openstack.org/372573 | 04:21 |
*** mingyu has quit IRC | 04:23 | |
*** dmorita has joined #openstack-swift | 04:27 | |
*** abhitechie has joined #openstack-swift | 04:30 | |
*** dmorita has quit IRC | 04:32 | |
*** SkyRocknRoll has joined #openstack-swift | 04:37 | |
lhx_ | mattoliverau, hi, could you spare some minutes for review? https://review.openstack.org/#/c/378203/ | 04:41 |
patchbot | patch 378203 - swift - Use six.moves.urllib.parse instead of urllib | 04:41 |
*** tongli has joined #openstack-swift | 04:52 | |
*** tongli_ has quit IRC | 04:55 | |
*** psachin has joined #openstack-swift | 04:57 | |
mattoliverau | lhx_: Sure, thanks for pointing me at the other change. That's going in now, so yours now should be good to :) | 05:02 |
*** manous has quit IRC | 05:03 | |
lhx_ | mattoliverau, :) | 05:07 |
*** ppai has joined #openstack-swift | 05:08 | |
lhx_ | mattoliverau, haha, looking forward to seeing | 05:11 |
*** abhinavtechie has joined #openstack-swift | 05:13 | |
*** abhitechie has quit IRC | 05:13 | |
*** mingyu has joined #openstack-swift | 05:14 | |
*** manous has joined #openstack-swift | 05:16 | |
*** manous has quit IRC | 05:23 | |
*** manous has joined #openstack-swift | 05:35 | |
*** dmorita has joined #openstack-swift | 05:37 | |
openstackgerrit | Merged openstack/swift: Py3: Fixes eventlet ImportError: No module named urllib2 https://review.openstack.org/348272 | 05:39 |
*** dmorita has quit IRC | 05:42 | |
*** bjolo has joined #openstack-swift | 05:54 | |
*** rcernin has joined #openstack-swift | 06:07 | |
*** niknakpaddywak has quit IRC | 06:21 | |
*** abhinavtechie has quit IRC | 06:24 | |
*** abhinavtechie has joined #openstack-swift | 06:25 | |
*** niknakpaddywak has joined #openstack-swift | 06:28 | |
*** pcaruana has joined #openstack-swift | 06:39 | |
*** sgundur has joined #openstack-swift | 06:45 | |
*** hseipp has joined #openstack-swift | 06:48 | |
*** sgundur has quit IRC | 06:49 | |
*** klrmn has quit IRC | 07:00 | |
*** manous has quit IRC | 07:05 | |
*** abhinavtechie has quit IRC | 07:15 | |
*** manous has joined #openstack-swift | 07:17 | |
*** rledisez has joined #openstack-swift | 07:20 | |
*** geaaru has joined #openstack-swift | 07:22 | |
*** jamielennox is now known as jamielennox|away | 07:23 | |
*** abhitechie has joined #openstack-swift | 07:25 | |
*** sgundur has joined #openstack-swift | 07:57 | |
*** jordanP has joined #openstack-swift | 07:58 | |
*** manous has quit IRC | 07:58 | |
*** sgundur has quit IRC | 08:02 | |
*** mingyu has quit IRC | 08:08 | |
*** Jeffrey4l_ has joined #openstack-swift | 08:09 | |
*** manous has joined #openstack-swift | 08:10 | |
*** Jeffrey4l has quit IRC | 08:10 | |
*** mingyu has joined #openstack-swift | 08:10 | |
openstackgerrit | Merged openstack/swift: Use six.moves.urllib.parse instead of urllib https://review.openstack.org/378203 | 08:15 |
*** acoles_ is now known as acoles | 08:18 | |
*** mingyu has quit IRC | 08:22 | |
*** niknakpaddywak has quit IRC | 08:25 | |
*** niknakpaddywak has joined #openstack-swift | 08:26 | |
acoles | kota_: hi, could you check this out https://review.openstack.org/379649 thanks | 08:32 |
patchbot | patch 379649 - swift - Fix failing cross-policy-versioning functional test | 08:32 |
acoles | notmyname: I think that will need backporting to the latest release ^^ | 08:32 |
kota_ | acoles: sure and good morning for you. | 08:32 |
acoles | kota_: thanks! | 08:32 |
acoles | friday aya | 08:33 |
acoles | yay* duh | 08:33 |
acoles | oh so the reconstructor mem leak bug was down to the pyeclib thing that kota already fixed. nice. | 08:33 |
*** manous has quit IRC | 08:34 | |
kota_ | acoles: If i understand it correctly ;-) | 08:34 |
acoles | i'm sure you do! | 08:34 |
kota_ | acoles: perhaps, it comes from calling policy.fragment_size which calls get_fragment_info at pyeclib that causes memory leak <- I fixed! | 08:35 |
acoles | kota_: and that would be consistent with the leak only occurring for sync type reconstructor jobs, the revert jobs don't use pyeclib | 08:36 |
kota_ | acoles: exactly. | 08:37 |
acoles | of course sync jobs do a bunch of other extra work too, but it is at least a plausible cause | 08:37 |
rledisez | acoles, kota_: i'll give a try today to this new pyeclib version, on a proxy and an object server | 08:38 |
acoles | rledisez: great! | 08:38 |
*** SkyRocknRoll has quit IRC | 08:42 | |
*** manous has joined #openstack-swift | 08:45 | |
openstackgerrit | zheng yin proposed openstack/swift: fix word spelling mistake https://review.openstack.org/379909 | 08:46 |
kota_ | acoles: gotcha on patch 379649 and looks good catch | 08:48 |
patchbot | https://review.openstack.org/#/c/379649/ - swift - Fix failing cross-policy-versioning functional test | 08:48 |
kota_ | acoles: but i have an idea to improve the patch more, will add an comment for that | 08:49 |
acoles | kota_: are you going to suggest cross policy tests with both modes? | 08:56 |
kota_ | acoles: no, just suggesting to use the variable | 08:57 |
kota_ | to keep consitency | 08:57 |
acoles | oic. :) | 08:57 |
*** SkyRocknRoll has joined #openstack-swift | 08:57 | |
kota_ | acoles: but it sounds cool ;-) | 08:57 |
acoles | It did occur to me but then I decided to just fix the broken tests quickly and go home ') | 08:58 |
acoles | adding more tests would be another patch IMO | 08:58 |
kota_ | I'm with :D | 08:59 |
*** lhx_ has quit IRC | 09:02 | |
*** lhx_ has joined #openstack-swift | 09:03 | |
openstackgerrit | Alistair Coles proposed openstack/swift: Make tempurl functional tests clean up account keys https://review.openstack.org/346752 | 09:06 |
acoles | kota_: good idea, I will make wishlist bug to look at making one of the in process functional test run with two policies (if that is possible) | 09:09 |
kota_ | acoles: super cool! | 09:10 |
*** b00tcat has left #openstack-swift | 09:16 | |
*** cbartz has joined #openstack-swift | 09:18 | |
openstackgerrit | Alistair Coles proposed openstack/swift: Fix failing cross-policy-versioning functional test https://review.openstack.org/379649 | 09:20 |
acoles | kota_: ^^ | 09:21 |
kota_ | acoles: let me try | 09:23 |
acoles | notmyname: ignore my earlier remark about backport, the bug landed after the release | 09:25 |
openstackgerrit | zheng yin proposed openstack/swift: Add test cases in the common/test_db https://review.openstack.org/366689 | 09:30 |
*** manous has quit IRC | 09:32 | |
*** manous has joined #openstack-swift | 09:45 | |
*** lhx_ has quit IRC | 09:47 | |
*** kei_yama has quit IRC | 09:54 | |
*** dmorita has joined #openstack-swift | 09:55 | |
*** dmorita has quit IRC | 10:00 | |
*** sgundur has joined #openstack-swift | 10:00 | |
*** sgundur has quit IRC | 10:05 | |
*** lhx_ has joined #openstack-swift | 10:20 | |
*** mvk has quit IRC | 10:30 | |
*** zhengyin has quit IRC | 10:34 | |
*** admin6 has left #openstack-swift | 10:35 | |
*** delattec has quit IRC | 10:36 | |
openstackgerrit | Thiago da Silva proposed openstack/swift: Symlink implementation. https://review.openstack.org/232162 | 10:42 |
*** manous has quit IRC | 10:46 | |
openstackgerrit | Kazuhiro MIYAHARA proposed openstack/swift: Use __init__() of Daemon class in sub classes https://review.openstack.org/365614 | 10:50 |
rledisez | acoles, kota_: i tried pyeclib 1.3 on a proxy. it seems there is memory consumption reduction. each worker still grow to more than a 1GB of RAM each hour but i think it's related to socket leak (we reload proxies every hours to free memory and sockets, more than one hour between each reload does not make it with 256GB of RAM) | 10:55 |
rledisez | i can't really test for the reconstructor as i wrote a workaround for https://bugs.launchpad.net/swift/+bug/1469815 which mitigate the memleak problem | 10:55 |
openstack | Launchpad bug 1469815 in OpenStack Object Storage (swift) "If the fragment needing to be reconstructed already exists, let the handoff node push it to the primary." [Low,In progress] - Assigned to Minwoo Bae (minwoob) | 10:55 |
rledisez | dewanee, admin6: | 10:57 |
rledisez | http://web.ledisez.net/reconstructor-1.patch | 10:57 |
rledisez | allows to start a reconstructor process per device, per job type, can put many devices using comma separator | 10:57 |
rledisez | Revert on first disk: /opt/swift/bin/swift-object-reconstructor /etc/swift/object-server.conf -r -d disk-00-000 | 10:57 |
rledisez | Sync on 2nd and 3rd disk: /opt/swift/bin/swift-object-reconstructor /etc/swift/object-server.conf -s -d disk-00-001,disk-00-002 | 10:57 |
rledisez | also it randomizes partitions order so you don't restart from beggining when restarting process | 10:57 |
rledisez | this patch inspired https://review.openstack.org/#/c/340584/, unit test were not passing on the gate because i would had to insert a sleep(0) to trigger a "thread switching" for the logging thread to run as the unit test relies on the log output. i didn't waste time on this | 10:57 |
patchbot | patch 340584 - swift - Make handoffs_first per partition in reconstructor | 10:57 |
rledisez | http://web.ledisez.net/reconstructor-2.patch | 10:57 |
rledisez | it's a fix for https://bugs.launchpad.net/swift/+bug/1469815 | 10:57 |
openstack | Launchpad bug 1469815 in OpenStack Object Storage (swift) "If the fragment needing to be reconstructed already exists, let the handoff node push it to the primary." [Low,In progress] - Assigned to Minwoo Bae (minwoob) | 10:57 |
rledisez | you really want it as it will save a lot of IO on your clusters | 10:57 |
rledisez | and it actually mitigates the memleak problem because there is SYNC only when really needed, not when rebalancing | 10:57 |
rledisez | for this patch to work you need to distribute the builder with the ring.gz | 10:57 |
*** manous has joined #openstack-swift | 10:58 | |
*** mvk has joined #openstack-swift | 11:00 | |
*** admin6 has joined #openstack-swift | 11:02 | |
admin6 | c | 11:02 |
*** manous has quit IRC | 11:03 | |
*** kmARC has quit IRC | 11:10 | |
*** kmARC has joined #openstack-swift | 11:11 | |
*** manous has joined #openstack-swift | 11:15 | |
*** tdasilva_ has quit IRC | 11:18 | |
*** bjolo has quit IRC | 11:27 | |
*** bjolo has joined #openstack-swift | 11:27 | |
*** portante has quit IRC | 11:50 | |
*** ndk_ has quit IRC | 11:50 | |
*** cdelatte has joined #openstack-swift | 11:56 | |
*** ndk_ has joined #openstack-swift | 11:59 | |
*** nikivi has joined #openstack-swift | 11:59 | |
*** portante has joined #openstack-swift | 11:59 | |
*** nikivi has quit IRC | 11:59 | |
*** ndk_ has quit IRC | 12:03 | |
*** ndk_ has joined #openstack-swift | 12:04 | |
*** portante has quit IRC | 12:04 | |
*** portante has joined #openstack-swift | 12:10 | |
*** amoralej is now known as amoralej|lunch | 12:14 | |
*** dfflanders has quit IRC | 12:23 | |
dewanee | thanks rledisez | 12:29 |
*** geaaru has quit IRC | 12:39 | |
admin6 | dewanee: rledisez: I missed the last 3 hours of IRC. rledisez. did you post the patch somewhere? | 12:42 |
rledisez | admin6: yes, you can find the logs here: http://eavesdrop.openstack.org/irclogs/%23openstack-swift/ | 12:47 |
*** Jeffrey4l_ has quit IRC | 12:47 | |
*** Jeffrey4l_ has joined #openstack-swift | 12:48 | |
*** psachin has quit IRC | 12:50 | |
admin6 | rledisez: thanks, one more thing I learned today :-) | 12:50 |
*** geaaru has joined #openstack-swift | 12:51 | |
*** SkyRocknRoll has quit IRC | 12:53 | |
*** links has quit IRC | 12:56 | |
*** david-lyle has joined #openstack-swift | 12:57 | |
*** mingyu has joined #openstack-swift | 13:05 | |
*** mingyu has quit IRC | 13:10 | |
openstackgerrit | Yushiro FURUKAWA proposed openstack/swift: Revises 'url' to 'URL' and 'json' to 'JSON' https://review.openstack.org/379390 | 13:22 |
*** amoralej|lunch is now known as amoralej | 13:22 | |
*** ppai has quit IRC | 13:40 | |
openstackgerrit | Alistair Coles proposed openstack/swift: EC - eliminate .durable files https://review.openstack.org/355958 | 13:49 |
*** mingyu has joined #openstack-swift | 13:55 | |
*** abhitechie has quit IRC | 14:05 | |
*** nikivi has joined #openstack-swift | 14:06 | |
*** vint_bra has joined #openstack-swift | 14:09 | |
*** antwash has left #openstack-swift | 14:13 | |
*** nikivi has quit IRC | 14:20 | |
*** bjolo has quit IRC | 14:21 | |
*** Jeffrey4l_ has quit IRC | 14:22 | |
*** Jeffrey4l_ has joined #openstack-swift | 14:23 | |
*** nikivi has joined #openstack-swift | 14:36 | |
*** manous has quit IRC | 14:36 | |
*** david-lyle has quit IRC | 14:37 | |
*** david-lyle has joined #openstack-swift | 14:38 | |
*** nikivi has quit IRC | 14:39 | |
openstackgerrit | ChangBo Guo(gcb) proposed openstack/swift: Add unit test content https://review.openstack.org/353444 | 14:51 |
openstackgerrit | ChangBo Guo(gcb) proposed openstack/swift: Add test_long_names in the test_account and test_container https://review.openstack.org/368627 | 14:51 |
*** acoles is now known as acoles_ | 14:51 | |
openstackgerrit | ChangBo Guo(gcb) proposed openstack/swift: simplify code https://review.openstack.org/355858 | 14:52 |
*** dmorita has joined #openstack-swift | 14:55 | |
*** sgundur has joined #openstack-swift | 14:58 | |
*** jaakkos has quit IRC | 14:58 | |
*** jaakkos has joined #openstack-swift | 14:58 | |
*** sgundur has quit IRC | 14:59 | |
*** dmorita has quit IRC | 14:59 | |
openstackgerrit | Merged openstack/swift: Improve functional tests and test client https://review.openstack.org/376455 | 15:06 |
*** jordanP has quit IRC | 15:24 | |
*** ChanServ sets mode: +v tdasilva | 15:24 | |
*** klrmn has joined #openstack-swift | 15:24 | |
*** nadeem has joined #openstack-swift | 15:25 | |
*** sgundur has joined #openstack-swift | 15:26 | |
*** nadeem has quit IRC | 15:26 | |
*** nadeem has joined #openstack-swift | 15:27 | |
*** hseipp has quit IRC | 15:28 | |
notmyname | good morning | 15:31 |
*** acoles_ is now known as acoles | 15:34 | |
*** chsc has joined #openstack-swift | 15:35 | |
*** chsc has joined #openstack-swift | 15:35 | |
*** klrmn has quit IRC | 15:35 | |
notmyname | rledisez: thanks for publishing that patch. I definitely want to associate it with the bug and see what we can get landed upstream | 15:37 |
*** silor has joined #openstack-swift | 15:38 | |
*** mvk has quit IRC | 15:39 | |
notmyname | tdasilva: timburke: kota_: saw some scrollback in a different channel about the pyeclib release. not sure what's going on or if there are any problems. but I asked fungi to come in here and help explain | 15:40 |
notmyname | acoles: did you look at rledisez's patch for EC? | 15:41 |
tdasilva | notmyname: was it in #openstack-infra? | 15:42 |
notmyname | no, one of the meeting channels | 15:43 |
*** fungi has joined #openstack-swift | 15:46 | |
notmyname | hi fungi | 15:46 |
notmyname | thanks for dropping by | 15:46 |
fungi | howdy! for some reason i realized i hadn't put this on autojoin | 15:47 |
notmyname | fungi: I wanted to ask you about the pyeclib thing you were talking about. and in here so kota_ timburke and tdasilva could all see it too (despite tz differences) | 15:47 |
notmyname | so, what's up with pyeclib? | 15:47 |
fungi | excellent--i wanted to hunt someone down about it | 15:47 |
notmyname | this is the place | 15:47 |
notmyname | I tagged a release yesterday for it | 15:48 |
notmyname | but it didn't get pushed to pypi | 15:48 |
fungi | sdists weren't built properly for their releases, i'm just now getting back to analyzing the log from it | 15:48 |
notmyname | but it's on tarbals.o.o | 15:48 |
fungi | most likely reason is an invalid trove classifier. pypi will reject uploads with metadata it doesn't recognize | 15:48 |
notmyname | hmm | 15:48 |
notmyname | I didn't change any of that since the last 1.2.1 release | 15:49 |
notmyname | which is on pypi | 15:49 |
fungi | aha, no it looks like we have an assumption baked into the signing script that it will be able to parse the package name out of setup.cfg | 15:50 |
fungi | we probably just need a fallback in that, and then i can retrigger the job | 15:50 |
fungi | i'm looking real quick to see how we should go about inferring package names (e.g. from the repo name perhaps) | 15:51 |
fungi | for situations where there's no setup.cfg in the repo | 15:51 |
clarkb | also we want to not make a wheel for it I think | 15:51 |
clarkb | since it links against liberasurecode | 15:52 |
notmyname | fungi: FWIW, the history is that pyeclib was written (and used) outside of openstack, used by swift most prominently, original authors don't have a lot of time, so we moved it under openstack/* to allow corps to contribute to it. but it's not "official" because C, and the TC seems hostile to that. so it hasn't been pbr-ified or anything. (same story for liberasurecode) | 15:52 |
fungi | clarkb: is linking the wheel against liberasurecode an issue? | 15:53 |
*** cbartz has left #openstack-swift | 15:53 | |
fungi | do we obtain it from somewhere weird or something? | 15:53 |
fungi | not distro packages? | 15:53 |
onovy | distro packages is here | 15:53 |
onovy | (debian) | 15:53 |
clarkb | fungi: it is becuse wheels have no way to specifythat dep | 15:53 |
clarkb | so when someone installs that whrel.chances are it willbe broken for them | 15:53 |
clarkb | unless on the same distro as our wheelbuilder | 15:54 |
fungi | clarkb: well, that's the case for a lot of wheels on pypi though | 15:54 |
clarkb | we want to force them to build from an sdist and do linking on their side | 15:54 |
fungi | oh, i get you | 15:54 |
fungi | we should avoid building and uploading wheels that aren't pure python | 15:54 |
fungi | right | 15:54 |
clarkb | yup | 15:54 |
clarkb | sdist is fine | 15:54 |
fungi | unless we get onboard with the rarified linux built set or whatever it is for that pep | 15:55 |
fungi | s/built/build/ | 15:55 |
clarkb | ya which liberasure code isnt part of | 15:55 |
clarkb | that env is very limited | 15:55 |
fungi | "manylinux" that's what it was called | 15:55 |
fungi | right-o | 15:55 |
fungi | okay, so i believe we do have a job that only builds sdists and no wheels we can swap into the layout for this repo | 15:56 |
fungi | i'll look at that once i get the signing script patched to guess a distname when there's no setup.cfg | 15:56 |
notmyname | please let me know what we need to do on the pyeclib side | 15:57 |
fungi | also, we aren't uploading signatures to pypi (since that feature is going away) so we should be able to make the pypi upload not dependind on the signing job | 15:58 |
fungi | ahh, it's already that way | 15:58 |
fungi | so the signing job failure shouldn't have prevented the pypi upload. checking the other log now | 15:58 |
fungi | aha! | 15:59 |
fungi | the sdist builds a tarball with a mixed-case name | 16:00 |
fungi | and since there's no setup.cfg, the pypi upload job doesn't know how to find the proper distname so it falls back to trying to base it on the distro name | 16:00 |
fungi | er, base it on the repo name | 16:00 |
notmyname | does that come from https://github.com/openstack/pyeclib/blob/master/setup.py#L174 ? | 16:00 |
acoles | notmyname: I haven't had chance to study rledisez's patch yet but I did bookmark it | 16:00 |
notmyname | acoles: thanks | 16:00 |
fungi | compare https://tarballs.openstack.org/pyeclib/ with http://git.openstack.org/cgit/openstack/pyeclib/tree/ | 16:01 |
fungi | er, with http://logs.openstack.org/c0/c0010b425247fa75a0c69c9201629d8ad8c635c3/release/pyeclib-pypi-both-upload/4914907/console.html | 16:01 |
notmyname | ya, the 404s | 16:01 |
fungi | we're looking for https://tarballs.openstack.org/pyeclib/pyeclib-1.3.0.tar.gz and the job doesn't know that pyeclib builds https://tarballs.openstack.org/pyeclib/PyECLib-1.3.0.tar.gz | 16:01 |
fungi | notmyname: yeah, that's the setup(name='PyECLib', in setup.py | 16:02 |
notmyname | fungi: that's where the mixed case comes from? | 16:02 |
acoles | notmyname: I'm OOO Monday but will try to take a look at it next week | 16:02 |
fungi | notmyname: right | 16:03 |
fungi | basically since the upload node has our project credentials for pypi on it, we can't safely execute setup.py for projects to find out their distro name spelling/case | 16:03 |
notmyname | ok | 16:04 |
notmyname | anything you'd worry about breaking by changing that name? | 16:04 |
fungi | normally we look it up in setup.cfg, which most of our packages use because of pbr needing it, and then the fallback for non-pbr-using packages is to assume the dist name matches teh repo name | 16:04 |
fungi | notmyname: it shouldn't break anything at all. pypi already normalizes all upper case to lower case and all punctuation to - | 16:04 |
notmyname | ack | 16:05 |
fungi | so it's a cosmetic change | 16:05 |
fungi | however, since it's a change to the repo, it'll need a new release tagged with that merged | 16:05 |
fungi | looking now at what we need to do to the layout to only build/upload sdists and not wheels | 16:06 |
notmyname | ya. working on that now | 16:06 |
notmyname | and because it's not using pbr, I can't use the 1 feature of pbr that I like: tag-based versioning :-( | 16:07 |
*** rcernin has quit IRC | 16:08 | |
fungi | hrm, actually the versions we have are for uploading only wheels (from the old prereleases are wheel-only days) and uploading both sdists and wheels. the tarball job also seems to build both tarballs and wheels. this will probably need a little deeper surgery | 16:08 |
fungi | notmyname: anyway, to test that the tarball will end up lower-case, before you tag it, run `tox -e venv python setup.py sdist` and then look at the tarball it creates | 16:10 |
notmyname | fungi: does that look right? https://review.openstack.org/380424 | 16:10 |
patchbot | patch 380424 - pyeclib - Updated name in setup.py to work with release tool... | 16:10 |
notmyname | hmm...bot didn't report that. is openstackbot ok? | 16:10 |
fungi | patchbot reported it | 16:10 |
notmyname | because I pasted the URL. normally it reports via openstackbot like normal | 16:11 |
notmyname | anyway...separate issue | 16:11 |
notmyname | fungi: pulling up my VM now to build the sdist | 16:11 |
fungi | yeah, looks right to me | 16:12 |
fungi | yeah, i don't have liberasurecode on my distro apparently | 16:12 |
notmyname | fungi: "venv: commands succeeded" | 16:13 |
fungi | aha, -dev | 16:14 |
fungi | notmyname: look in the new dist/ directory | 16:14 |
fungi | i see pyeclib-1.3.1.tar.gz now | 16:14 |
notmyname | $ ls | 16:14 |
notmyname | pyeclib-1.3.1.tar.gz | 16:14 |
notmyname | same here | 16:14 |
notmyname | so that's what's supposed to be there? | 16:14 |
fungi | which looks like how the pypi upload job expected to find the tarball | 16:14 |
fungi | yep | 16:15 |
notmyname | anything else to do before tagging, then? | 16:15 |
fungi | without your patch it was creating PyECLib-1.3.0.tar.gz | 16:15 |
*** rledisez has quit IRC | 16:16 | |
notmyname | tdasilva: you follow? ^ | 16:16 |
*** sgundur has quit IRC | 16:16 | |
fungi | you should be safe to tag. clarkb's concern about the wheels should be less urgent since the upload job isn't going to find it anyway with -cp27-cp27mu-linux_x86_64 for the architecture | 16:16 |
clarkb | that may cause the job to fail though | 16:17 |
fungi | clarkb: no, we have lots of projects that don't build wheels at all, so the upload script knows to just skip wheel uploads if it can't find one | 16:17 |
notmyname | I can hold off on panding that patch and tagging, if needed | 16:17 |
tdasilva | notmyname: following, just wondering if we do need 1.3.1? | 16:18 |
fungi | and it won't find these because they're named with an arch it won't know to look for | 16:18 |
tdasilva | can't we just re-tag 1.3.0? is that really bad? | 16:18 |
tdasilva | especially since it didn't make to pypi | 16:18 |
fungi | tdasilva: yeah, deletion/replacement of git tags doesn't propagate except via push | 16:18 |
tdasilva | not that it's a big deal | 16:18 |
notmyname | tdasilva: code change upstream means we should bump the rev number | 16:18 |
*** mingyu has quit IRC | 16:18 | |
fungi | so people who already pulled the old tag (including our release automation) won't know to pull the replacement tag if it has the same tag name | 16:19 |
tdasilva | fungi, notmyname : ok | 16:19 |
notmyname | torgomatic keeps telling me we won't run out of version numbers ;-) | 16:20 |
fungi | yep, and in general reusing tags in a public git repo is bad hygiene | 16:20 |
*** hseipp has joined #openstack-swift | 16:21 | |
*** sgundur has joined #openstack-swift | 16:22 | |
fungi | also, our signing job already had the same fallback i was talking about adding, i just didn't read down far enough. it reuses the same retrieval logic as the pypi upload job. so the name lower-casing should now cause both of those jobs to work | 16:23 |
notmyname | clarkb: are you ok with me landing the patch and tagging, or is there other stuff that needs to happen first? | 16:23 |
fungi | (meaning we'll also see a detached signature for the 1.3.1 sdist appear on tarballs.o.o) | 16:23 |
clarkb | notmyname: if fungi is happy then I am too | 16:23 |
notmyname | :-) | 16:23 |
notmyname | words to live by? | 16:23 |
clarkb | indeed | 16:24 |
clarkb | but also I just walkes into an openstackday (seattle) so distracted | 16:24 |
fungi | clarkb: yeah, i think in the long term we need an sdist-only version of the tarball job, but for now the wheels for this are only "published" to tarballs.o.o (and won't even have a signature with them) | 16:24 |
fungi | we're sort of saved by the fact that our jobs assume you either have a pure python wheel or no wheel at all | 16:25 |
notmyname | fungi: clarkb: thanks for your help this morning | 16:25 |
fungi | notmyname: happy to help! sorry this took you by surprise, and took as much time to track down | 16:25 |
notmyname | no worries | 16:26 |
fungi | i have a feeling previous releases were manually uploaded, because i don't see any earlier tarballs on tarballs.o.o | 16:26 |
openstackgerrit | Merged openstack/pyeclib: Updated name in setup.py to work with release tooling. https://review.openstack.org/380424 | 16:26 |
tdasilva | notmyname: we should get fungi some beer or sangria in barca ;) | 16:27 |
fungi | we don't make a habit of removing files from there, and our pypi upload automation (as you've seen) relies on them being there | 16:27 |
fungi | tdasilva: i have a feeling there will be no shortage, but thanks | 16:27 |
fungi | i'm always happy to grab a drink with such awesome colleagues | 16:28 |
notmyname | done. patch landed. tag pushed | 16:28 |
notmyname | post-tag job should be in zuul now | 16:29 |
tdasilva | fungi: thanks for your help :) | 16:29 |
fungi | release jobs are running for it now | 16:30 |
fungi | it's on tarballs.o.o now, looks right so far | 16:31 |
fungi | found the right tarball and is uploading to pypi | 16:31 |
notmyname | and there it is https://pypi.python.org/pypi/PyECLib | 16:31 |
fungi | yep | 16:32 |
fungi | also there's a pyeclib-1.3.1.tar.gz.asc (detached openpgp signature) showing on https://tarballs.openstack.org/pyeclib/ now | 16:32 |
fungi | so looks like the stuff we expected to work worked this time | 16:33 |
fungi | don't hesitate to let me or someone in #openstack-infra know if you spot any other issues you think might be related, but i believe you're all set now | 16:33 |
notmyname | tdasilva: we should bump global requirements to have >=1.3.1 | 16:34 |
tdasilva | why did all the text from the README disappear from https://pypi.python.org/pypi/PyECLib ? | 16:35 |
*** sgundur has quit IRC | 16:36 | |
*** sgundur has joined #openstack-swift | 16:37 | |
*** dmorita has joined #openstack-swift | 16:40 | |
tdasilva | notmyname: are you planning on making a new cut for newton? | 16:44 |
tdasilva | with the 1.3.1 requirement? | 16:44 |
*** sgundur has quit IRC | 16:44 | |
notmyname | no | 16:58 |
notmyname | and no idea on why the README isn't seen any more | 16:58 |
*** klrmn has joined #openstack-swift | 17:01 | |
*** acoles is now known as acoles_ | 17:05 | |
fungi | the readme needs to be included in the metadata at upload | 17:10 |
fungi | i'm familiar with how we configure that in setup.cfg, there's probably some analog in straight setup.py structures | 17:10 |
fungi | in setup.cfg we set, e.g., metadata.description-file = README.rst | 17:11 |
*** nadeem has quit IRC | 17:11 | |
fungi | looks like that's semi-automated by pbr (sucking a file's contents into the description parameter), and the pypa example shows implementing it in setup.py like this: https://github.com/pypa/sampleproject/blob/master/setup.py#L16 | 17:14 |
fungi | tdasilva: notmyname: ^ | 17:14 |
*** nadeem has joined #openstack-swift | 17:15 | |
fungi | basically extract the content from the readme file and then pass it as a string in the long_description parameter to setup() | 17:15 |
fungi | should be fairly straightforward | 17:16 |
*** mingyu has joined #openstack-swift | 17:19 | |
*** mvk has joined #openstack-swift | 17:23 | |
*** mingyu has quit IRC | 17:24 | |
*** amoralej is now known as amoralej|off | 17:31 | |
*** tqtran has joined #openstack-swift | 17:32 | |
*** tongli has quit IRC | 17:47 | |
notmyname | tdasilva: how can we track updating the README in pyeclib? ie I'm not going to work on that right at this minute, but it's somethat needs to be done? bug in LP? any other dieas? | 17:50 |
notmyname | tdasilva: https://bugs.launchpad.net/pyeclib/+bug/1629407 | 17:54 |
openstack | Launchpad bug 1629407 in PyECLib "fix readme for pypi publishing" [Undecided,New] | 17:54 |
*** nadeem has quit IRC | 17:57 | |
*** lhx_ has quit IRC | 17:59 | |
notmyname | https://bugs.launchpad.net/swift/+bug/1629410 <-- update dependency version | 18:03 |
openstack | Launchpad bug 1629410 in OpenStack Object Storage (swift) "update pyeclib dependency to >=1.3.1" [Medium,New] | 18:03 |
*** geaaru has quit IRC | 18:04 | |
*** sgundur has joined #openstack-swift | 18:12 | |
*** nikivi has joined #openstack-swift | 18:17 | |
*** nikivi has quit IRC | 18:18 | |
*** hseipp has quit IRC | 18:18 | |
*** nikivi has joined #openstack-swift | 18:19 | |
*** hseipp has joined #openstack-swift | 18:19 | |
*** hseipp has quit IRC | 18:19 | |
*** nikivi has quit IRC | 18:19 | |
tdasilva | notmyname: assigned the pyeclib to me | 18:23 |
tdasilva | notmyname: i'm also wondering if we should go ahead and create a setup.cfg file? and organize things more like we have in swift | 18:24 |
tdasilva | ? | 18:24 |
notmyname | it's not a bad idea | 18:31 |
* notmyname needs to step about a bit | 18:32 | |
*** arch-nemesis has joined #openstack-swift | 18:40 | |
*** MVenesio has joined #openstack-swift | 18:50 | |
*** aswadr_ has quit IRC | 18:52 | |
* MooingLemur drops a bit for stepping about. | 18:58 | |
clayg | lol | 19:00 |
clayg | am I missing something? this line is duplicated bs right -> https://github.com/openstack/swift/blob/29516db36e2707fe195db5fc13d07e801d92e5b2/swift/common/ring/utils.py#L627 | 19:12 |
timburke | clayg: seems to be | 19:13 |
*** openstackgerrit has quit IRC | 19:18 | |
*** openstackgerrit has joined #openstack-swift | 19:19 | |
openstackgerrit | Clay Gerrard proposed openstack/swift: Remove unneeded line from ring.utils.dispersion_report https://review.openstack.org/380518 | 19:19 |
*** cdelatte has quit IRC | 19:22 | |
*** cdelatte has joined #openstack-swift | 19:32 | |
clayg | i feel like I've never really been able to install swift on my mac since we added pyeclib - this makes it really annoying if you want to re-use code from swift's utils in other projects - e.g. common.ring :\ | 19:36 |
clayg | on one hand it's probably good since we don't really want swift utils to be consideres generally suitable to use as a library - OTOH, something like the builder/ring code is really useful in contexts outside of swift procsses | 19:37 |
*** sgundur has quit IRC | 19:37 | |
tdasilva | clayg: break swift in separate repos ;) | 19:38 |
clayg | e.g. the machine you run `swift-ring-builder` on has no expectation of being "in" the cluster beyond the fact that you can't get it installed anywhere that doesn't already have pyeclib on it | 19:38 |
clayg | I think this is a bit of a mistake/tragedy - un-intended consequence - maybe not the end of the world - but less than ideal | 19:38 |
clayg | tdasilva: your winky makes it seem like that's not the obvious solution | 19:38 |
clayg | tdasilva: another option is to try and make swift the python code "installable" without some dependencies - pyeclib as a depends could be moved into a 'swift-storage-policy-ec' plugin - it's less obvious what's appropriate for libffi/xattrs (some what easier to get installed with brew) | 19:40 |
tdasilva | clayg: of course i was just joking, but i do agree that it is unfortunate, although i never realized that until now. | 19:43 |
*** sgundur has joined #openstack-swift | 19:44 | |
*** admin6 has quit IRC | 19:44 | |
clayg | meh, i'm not sure my swift-ring-builder example is really that great - i think i'm mostly just being bitter about living with my own bad life choices - stupid apple. | 19:51 |
*** niknakpaddywak has quit IRC | 19:53 | |
*** tdasilva has quit IRC | 20:06 | |
*** dmorita has quit IRC | 20:27 | |
*** dmorita has joined #openstack-swift | 20:30 | |
*** Jeffrey4l_ has quit IRC | 20:35 | |
*** sgundur has quit IRC | 20:36 | |
*** sgundur has joined #openstack-swift | 20:37 | |
*** silor has quit IRC | 20:37 | |
*** tdasilva has joined #openstack-swift | 20:40 | |
*** sgundur has quit IRC | 20:43 | |
*** sgundur has joined #openstack-swift | 20:44 | |
*** MVenesio has quit IRC | 20:45 | |
*** mingyu has joined #openstack-swift | 20:57 | |
*** mingyu has quit IRC | 21:02 | |
openstackgerrit | Merged openstack/swift: Remove unneeded line from ring.utils.dispersion_report https://review.openstack.org/380518 | 21:03 |
*** cdelatte has quit IRC | 21:04 | |
clayg | how confused does the ring get if you say r1z1 and r2z1 - those are different zones (1, 1) and (2, 1) a device['zone']'s id is not expected to be unique? | 21:28 |
*** admin6 has joined #openstack-swift | 21:46 | |
*** nadeem has joined #openstack-swift | 21:53 | |
*** admin6 has quit IRC | 21:56 | |
*** jamielennox|away is now known as jamielennox | 21:59 | |
*** geaaru has joined #openstack-swift | 22:17 | |
*** vint_bra has quit IRC | 22:24 | |
*** sgundur has quit IRC | 22:29 | |
torgomatic | shouldn't get confused at all; (1, 1) and (2, 1) are different tuples | 22:34 |
*** arch-nemesis has quit IRC | 22:34 | |
torgomatic | it's confusing for humans, so you probably shouldn't do it ;) but I believe the ring code will work just fine | 22:34 |
*** geaaru has quit IRC | 22:40 | |
openstackgerrit | John Dickinson proposed openstack/swift: Update test-requirements.txt https://review.openstack.org/365271 | 22:46 |
openstackgerrit | Merged openstack/swift: Use ConfigParser instead of SafeConfigParser https://review.openstack.org/368617 | 22:52 |
*** nadeem has quit IRC | 23:18 | |
*** chsc has quit IRC | 23:27 | |
openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 23:28 |
*** sgundur has joined #openstack-swift | 23:44 | |
*** mingyu has joined #openstack-swift | 23:47 | |
*** furlongm has quit IRC | 23:50 | |
*** sgundur has quit IRC | 23:50 | |
*** sgundur has joined #openstack-swift | 23:51 | |
*** mingyu has quit IRC | 23:51 | |
*** furlongm has joined #openstack-swift | 23:52 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!