Thursday, 2010-09-23

*** joearnold has quit IRC00:01
*** pvo has quit IRC00:02
*** joearnold has joined #openstack00:02
*** ccustine has joined #openstack00:02
gundlacheday: i'm not sure what you're working on.  could you point me to a part of controllers-in-api that you're not doing and that is so broken it needs fixing before merging to trunk?00:03
*** joearnol_ has quit IRC00:03
gundlach[i think cloudpipe is off the list, right?]00:03
edaygundlach: I got cloudpipe in the works, as well as rpc.call00:08
edaytornado is still scattered through the test cases though, so it will be a bit more work to remove all that00:08
*** Cybo-mobile has joined #openstack00:08
mtayloreday: I need to come look through the new code you guys have been writing00:09
gundlachmtaylor: lp:~gundlach/nova/controllers-in-api, lp:~gundlach/nova/missing-ec2-url-endpoints00:10
gundlacheday: did somebody already fix the lingering nova.endpoint reference(s) in a branch, or shall I make one to do that?00:14
gundlach(i'm thinking it's maybe just nova-manage)00:15
edaygundlach: I have that in my branch already00:18
edaypart of the cloudpipe stuff00:19
gundlacheday: k; so is there anything besides the tornado test framework stuff?00:19
edaygundlach: not sure, anything else without unittests? :)00:20
gundlachheh, nothing that i'm aware of, but that doesn't say much!00:20
gundlacheday: i'm going to be around for a couple more hours to review&merge any branches you finish00:21
gundlachi was going to be mostly out Thu & Fri but I'll be in as needed to review & merge and get trunk back to normal00:22
*** pharkmillups has quit IRC00:22
*** Cybo-mobile has quit IRC00:22
*** sirp1 has quit IRC00:23
*** tobym has quit IRC00:29
*** lmcdowell has quit IRC00:29
edaygundlach: I can push up what I have, but cloudpipe is still broken while I still hunt down other bits00:30
gundlachk, fine with me -- i don't think cloudpipe has to be fixed to put this back in trunk (as it hasn't been blocking anybody since it broke with orm-deux afaik)00:30
*** jtdowney has quit IRC00:32
*** sophiap has quit IRC00:34
*** silassewell has quit IRC00:34
*** silassewell has joined #openstack00:35
*** ccustine has quit IRC00:49
*** scottie has joined #openstack00:53
edaygundlach: lp:~eday/nova/endpoint-cleanup is my work so far, still investigating some things and need to test run_instances properly00:56
gundlachchecking it out now00:56
gundlachfyi just proposed merging to controllers-in-api to start reviewing it00:57
*** DubLo7 has quit IRC01:00
*** DubLo7 has joined #openstack01:01
edaymmm, not quite ready yet... but ok :)01:02
*** joearnold has quit IRC01:06
* gundlach is *very* eager to get this back into trunk :)01:07
edayI gotta run, so feel free to improve/fix on my branch as you need01:08
edaywe can sync in the AM01:08
*** littleidea has quit IRC01:16
*** sophiap_ has joined #openstack01:16
*** kevnfx has joined #openstack01:17
*** hazmat has quit IRC01:18
*** adjohn has joined #openstack01:29
*** adjohn has quit IRC01:29
*** Ryan_Lane has quit IRC01:33
*** maplebed has quit IRC01:39
*** hazmat has joined #openstack01:43
*** hazmat has quit IRC01:52
*** pvo has joined #openstack02:03
*** ChanServ sets mode: +v pvo02:03
*** littleidea has joined #openstack02:05
*** ArdRigh has joined #openstack02:11
*** pvo has quit IRC02:12
*** sirp1 has joined #openstack02:12
*** pvo has joined #openstack02:12
*** pvo has joined #openstack02:12
*** ChanServ sets mode: +v pvo02:12
*** howardroark has quit IRC02:13
*** sirp1 has quit IRC02:26
*** DubLo7 has quit IRC02:31
*** DubLo7 has joined #openstack02:31
*** littleidea has quit IRC02:32
*** sirp1 has joined #openstack02:34
*** sirp1 has quit IRC02:42
*** sirp1 has joined #openstack02:42
*** gundlach has quit IRC02:45
*** howardroark has joined #openstack02:54
*** scottie has quit IRC02:55
*** gustavomzw has quit IRC02:58
deshantm_cosiinstalled nova using vishvananda's script when, running euca-run-instances -k test -t m1.tiny ami-tiny, we get "EC2_ACCESS_KEY environment variable must be set. Connection failed"02:59
*** howardroark has quit IRC03:01
*** howardroark has joined #openstack03:02
*** pvo has quit IRC03:19
*** pvo has joined #openstack03:19
*** ChanServ sets mode: +v pvo03:19
*** joearnold has joined #openstack03:20
*** littleidea has joined #openstack03:25
*** kashyapc has joined #openstack03:51
*** kashyapc has joined #openstack03:54
*** littleidea has quit IRC04:05
*** littleidea has joined #openstack04:09
*** kevnfx has quit IRC04:15
*** DubLo7 has quit IRC04:27
*** pvo_ has joined #openstack04:30
*** ChanServ sets mode: +v pvo_04:30
*** pvo has quit IRC04:33
*** pvo_ has quit IRC04:34
*** Ryan_Lane has joined #openstack04:35
*** guigui1 has joined #openstack04:37
*** joearnold has quit IRC04:55
*** pvo has joined #openstack04:58
*** ChanServ sets mode: +v pvo04:58
*** howardroark has quit IRC05:12
*** f4m8_ is now known as f4m805:14
*** f4m8 is now known as drehmoment05:18
*** abecc has quit IRC05:29
*** ccustine has joined #openstack05:30
*** rdw_ has joined #openstack05:42
*** silassewell1 has joined #openstack05:43
*** silassewell has quit IRC05:44
*** rdw_ has quit IRC05:50
*** rdw_ has joined #openstack05:51
*** rdw_ has quit IRC05:53
*** rdw_ has joined #openstack05:56
*** ccustine has quit IRC05:58
*** rdw has quit IRC05:59
*** rdw has joined #openstack05:59
*** rdw_ has quit IRC06:01
*** pvo has quit IRC06:01
*** rdw has joined #openstack06:02
*** silassewell1 has quit IRC06:04
*** ibarrera has joined #openstack06:15
*** zaitcev has quit IRC06:19
*** rlucio has joined #openstack06:23
*** rlucio has quit IRC06:38
*** gasbakid has joined #openstack06:39
*** allsystemsarego has joined #openstack06:41
*** allsystemsarego has joined #openstack06:41
*** mtaylor has quit IRC06:48
*** befreax has joined #openstack06:55
*** Ryan_Lane has quit IRC06:57
*** brd_from_italy has joined #openstack06:59
*** Ryan_Lane has joined #openstack07:01
*** Ryan_Lane has quit IRC07:10
*** krish has joined #openstack07:18
*** zheng_li has joined #openstack07:34
*** guigui1 has quit IRC08:01
*** sirp1 has quit IRC08:05
*** befreax has quit IRC08:05
*** befreax has joined #openstack08:08
*** befreax has quit IRC08:13
*** befreax has joined #openstack08:17
*** calavera has joined #openstack08:25
*** krish has quit IRC08:51
*** krish has joined #openstack08:51
soren_cerberus_: Fixed your unit tests.09:03
soren_cerberus_: I accept beer.09:03
soren_cerberus_: lp:~soren/nova/rs_auth_fixes ( has the fix.09:08
soren_cerberus_: To include it in your branch (not sure if you've done this before), just go into your checkout, do "bzr merge lp:~soren/nova/rs_auth_fixes; bzr commit -m 'Merge fix from Soren'; bzr push"09:09
soren_cerberus_: That's it.09:09
soren_cerberus_: I ran the test suite on the hudson box with that patch applied and everything works brilliantly.09:09
*** ar1 has quit IRC09:24
*** ar1 has joined #openstack09:25
*** littleidea has quit IRC09:46
*** gasbakid has quit IRC09:49
*** allsystemsarego has quit IRC10:01
*** gustavomzw has joined #openstack10:17
*** gustavomzw has quit IRC10:36
*** scottie has joined #openstack10:50
*** irahgel has joined #openstack10:52
*** dendrobates is now known as dendro-afk11:06
*** ArdRigh has quit IRC11:15
*** maplebed has joined #openstack11:27
*** ctennis has quit IRC11:33
*** gasbakid has joined #openstack11:58
sorenvishy: Ping me when you're around. I'm very interested in getting your libvirt problem sorted out.12:00
*** scottie has quit IRC12:01
*** scottie has joined #openstack12:09
*** krish has quit IRC12:09
*** scottie has quit IRC12:11
*** sophiap_ has quit IRC12:23
*** localhost has quit IRC12:26
*** localhost has joined #openstack12:27
*** sophiap has joined #openstack12:31
*** dendro-afk is now known as dendrobates12:40
*** gundlach has joined #openstack12:43
sorengundlach: Just the man I was looking for.12:47
gundlachsoren! oh good?12:47
sorengundlach: It appears your nova-api-new doesn't accept --flagfile.12:47
soren(and possible all other args as well)12:47
sorenI think I'm close to working out why.12:47
gundlachi could believe that.12:47
gundlacheday would probably be able to tell you more -- he wrote nova-api-new i believe12:48
sorenSo, nova/twistd takes care to pass sys.argv (after some twisted magic) to FLAGS12:48
sorengundlach: Oh, I thought that was all your fault :)12:48
sorennova/ does the same, I think.12:49
* soren checks12:49
gundlachsoren: i'm happy to take the blame, this is all part of the RS API blueprint :)12:49
sorenHmm... service doesn't.12:49
gundlachare you planning on adding that back in or would you like me to?  I'm doing some other work to get trunk right again, but can add that to the queue.12:49
sorengundlach: I'm working on it right now. I'll probably figure it out in a little bit.12:50
gundlachsoren: k12:50
gundlachsorry, and thanks12:50
sorenI can't move on until I've sorted it out, so I might as well.12:50
gundlachso you're working off of lp:~gundlach/nova/controllers-in-api, right?12:50
*** kashyapc has quit IRC12:50
sorenAh, nova/ does it.12:51
*** kashyapc has joined #openstack12:51
sorengundlach: I'm working off of what used to be the head of trunk.12:51
sorengundlach: because that's where I branched from.12:51
gundlachsoren: k.  no idea how bzr behaves under these conditions, but I bet if you have merge troubles mtaylor will be able to sort them out.12:51
sorengundlach: I know how it's going to behave. Removing stuff from an already published branch is just a really bad idea. Kinda like how rebasing a public git repo is a really bad idea.12:53
*** kashyapc has quit IRC12:54
sorengundlach: Do you happen to know how I tell it to daemonize?12:55
*** klord has joined #openstack12:56
gundlach'nova-api-new &' ?12:56
*** klord has quit IRC12:56
gundlachif you take a look at the code, you see it's basically 2 lines of code.12:56
*** klord has joined #openstack12:56
sorenI know.12:57
sorenso are all the other ones, yet they manage :)12:57
*** ppetraki has joined #openstack13:02
gundlachsoren: yeah, i'm not seeing anything in eventlet.wsgi.server docs about daemonizing.  I guess you'd have to write a wrapper if you don't want to use &.13:03
sorenI'm trying to use the wrapper the old api server used. I seems it should work, but I'm getting an odd error from somewhere inside eventlet :(13:04
*** gasbakid has quit IRC13:04
sorenI don't suppose that looks familiar?13:04
sorenthis is eventlet 0.9.12, fwiw.13:07
gundlachsoren: vish may have found the flagfile thing you were talking about -- see the merge thread for controllers-in-api into nova (the >70-email one)13:08
gundlachnear the bottom he talks puts in FLAGS(sys.argv) which sounds like what you mentioned13:08
sorenI tried that already.13:09
sorenThat's when I realised it didn't daemonize.13:09 now I'm trying to wrap it in nova.server.serve()13:09
gundlachsoren: no, i'm not familiar with the error paste13:10
*** ctennis has joined #openstack13:11
*** ctennis has joined #openstack13:11
sorenJust WOW!.13:12
sorenI tried googling for this error..13:13
sorenand Google gives me... My own pastebinned backtrace.13:13
sorenIt's been 8 minutes!13:13
gundlachgoogle is incorporating your browser history into search personalization, perhaps :)13:14
sorenYou would know.13:16
zulyeah i noticed that sometimes as well13:18
sorenoh, ffs.13:19
*** sonal has joined #openstack13:26
*** johnpur has joined #openstack13:29
*** ChanServ sets mode: +v johnpur13:29
*** tobym has joined #openstack13:30
* soren takes a break13:31
*** jdarcy has joined #openstack13:33
* soren headdesks13:33
sorengundlach: Fixed.13:33
gundlachsoren: which one13:33
gundlachgreat, send me a mergeprop to lp:~gundlach/nova/controllers-in-api13:34
sorengundlach: In a few minutes.13:35
*** kashyapc has joined #openstack13:35
sorenMy ps output looks much prettier now that there aren't eleventy billion nova-api processes.13:35
*** hazmat has joined #openstack13:39
*** sonal has quit IRC13:41
*** tobym has quit IRC13:42
*** drehmoment is now known as f4m8_13:43
sorengundlach: lp:~soren/nova/nova-api-new-daemonise13:47
sorengundlach: Er... still pushing.13:48
sorengundlach: There it goes.13:48
* soren kicks his DNS around a little bit.13:48
sorengundlach: You want an mp for it?13:50
gundlachsoren: sure13:50
*** befreax has quit IRC13:51
*** gasbakid has joined #openstack13:51
* soren goes back to trying to be useful :)13:52
gundlachcool, i'll review and merge.  lemme know what the next thing that gets in yoru way is13:52
gundlachhopefully we'll have trunk back to normal in a few hours13:52
sorenOh, I will. Trust me.13:52
*** littleidea has joined #openstack13:54
gundlachsoren: you proposed merging to trunk. can you propose to lp~gundlach/nova/controllers-in-api instead?  your diff is +283/-759 17 files modified.13:56
gundlach(or shall i just do it?)13:56
sorengundlach: Oh, whoops.13:56
*** Podilarius has joined #openstack13:56
*** scottie has joined #openstack13:57
*** dendrobates is now known as dendro-afk13:57
sorengundlach: There. All better:
*** joearnold has joined #openstack13:59
gundlachsoren: so utils.default_flagfile() can stay -- it doesn't need to become FLAGS(sys.argv), because server.serve handles that.  correct?13:59
sorengundlach: Right.14:00
*** littleidea has quit IRC14:00
gundlachdoes hudson do auto-merges into branches or only into trunk? (do i need to merge this myself)14:00
sorenOnly trunk.14:00
sorenlp:~gundlach/nova/controllers-in-api is only writable by you, you see.14:01
*** dendro-afk is now known as dendrobates14:01
gundlachdidn't know if LP let bots ignore that or something14:01
*** hazmat has quit IRC14:01
sorenThat would be very, very wrong. :)14:02
gundlachit wouldn't surprise me though ;)14:03
sorenThat's sad.14:03
gundlachbzr and LP are foreign enough to me, and keep having behaviors that i find unusual, to the point where i just go with the flow.  so if LP had a special case for Hudson to be able to merge approved mergeprops, hey, maybe that was useful and so somebody coded it in there.14:04
sorenbzr branches owned by you on Launchpad are yours and yours alone.14:05
*** joearnold has quit IRC14:05
gundlachsoren: ping -- any response?14:06
sorengundlach: Nope. I've reported it. Lovely timing.14:07
*** allsystemsarego has joined #openstack14:08
sorenIt's back.14:08
gundlachjust went out for coffee14:09
gundlachi imagine14:09
sorenIt might be in and out for a little bit:14:09
uvirtbotNew bug: #646031 in swift "COPY method doesn't support object names with slashes" [Medium,New]
sorengundlach: Next thing that doesn't work: nova-manage. It attempts to "from nova.endpoint import cloud" which fails miserably.14:13
gundlachsoren: that's been fixed in a branch somewhere...14:14
gundlachhold up14:14
* soren holds up14:14
gundlachright, lp:~eday/nova/endpoint-cleanup14:15
gundlachwhich he said was not complete when he left late last night, so i haven't yet merged it in.14:16
sorenAny reason not to merge that into your branch?14:16
uvirtbotNew bug: #646039 in swift "PUT with X-Copy-From header requires content length header" [Undecided,New]
sorenHeh :)14:16
gundlachi'll spot-fix that part and merge it and we'll see what else is blocking you.14:16
gundlachi think his branch may be larger than we care about for purposes of getting back into trunk and getting things normal again.14:16
sorenI'm not that blocked anymore. I can get stuff done now.14:16
gundlachok, so maybe it was you and vishy only.  once vishy is in we can check with him.14:16
*** tonywolf has joined #openstack14:19
gundlachsoren: bzr pull controllers-in-api14:20
sorenWhat did you do now? :)14:20
*** dendrobates is now known as dendro-afk14:21
*** tobym has joined #openstack14:23
gundlachsoren: just applied the endpoint fix you mentioned :)14:25
*** kevnfx has joined #openstack14:25
*** gasbakid has quit IRC14:30
sorenHmm... This is odd.14:30
sorenOh, not it's not.14:30
* soren headdesks14:31
sorenJust for good measure14:31
* soren has not yet gotten used to having to run a scheduler process.14:33
*** abecc has joined #openstack14:37
zulsoren: you are going to get concussed14:41
gundlachs'ok, he has a foam desktop just for this reason14:42
*** kevnfx has quit IRC14:43
sorenvishy: "good" news. I now see your libvirt problem. I'll try to get it sorted out.14:44
*** littleidea has joined #openstack14:48
*** pharkmillups has joined #openstack14:49
*** rnirmal has joined #openstack14:51
*** sirp1 has joined #openstack14:56
sorenvishy: Definitely an internal libvirt problem.14:56
*** tobym has quit IRC14:59
*** scottie has quit IRC15:03
*** zaitcev has joined #openstack15:04
*** kevnfx has joined #openstack15:19
*** calavera has quit IRC15:21
*** annegentle has joined #openstack15:23
*** blakeyeager has joined #openstack15:25
*** ibarrera has quit IRC15:29
*** cloudmeat1 has joined #openstack15:29
*** abien has quit IRC15:30
*** sirp1 has quit IRC15:31
*** cloudmeat has quit IRC15:32
* soren calls it a day15:42
sorenvishy: I'll work on that libvirt problem first thing tomorrow.15:42
*** scottie has joined #openstack15:44
*** brd_from_italy has quit IRC15:46
*** tonywolf has quit IRC15:49
*** spackest has joined #openstack15:49
*** krzycoder has quit IRC15:51
*** burris has quit IRC15:53
*** tonywolf has joined #openstack16:02
*** krish has joined #openstack16:03
*** dysinger has joined #openstack16:05
*** tonywolf has quit IRC16:05
*** Pentheus has joined #openstack16:13
kevnfxQ: what do I create a nova/compute/libvirt.xml.template file from?16:17
*** scottie has quit IRC16:18
*** scottie has joined #openstack16:20
*** zheng_li has quit IRC16:21
*** burris has joined #openstack16:23
*** mtaylor has joined #openstack16:29
*** ChanServ sets mode: +v mtaylor16:29
_cerberus_So the definition of insanity is performing the same action over and over and expecting different results. With that said, my branch continues to pass all tests locally and fail when Hudson runs them. Any suggestions?16:29
*** Ryan_Lane has joined #openstack16:31
*** tonywolf has joined #openstack16:31
*** kashyapc has quit IRC16:34
*** annegentle has quit IRC16:36
*** annegentle has joined #openstack16:37
*** gustavomzw has joined #openstack16:47
mtaylor_cerberus_: next thing I'd check would be library versions - of I can give you a login on the hudson machine so you can try debugging the issue there16:49
_cerberus_I'm using whatever the venv installs, so unless the file is out of date...16:50
*** zheng_li has joined #openstack16:52
mtaylorhrm. well, hudson does not use the venv16:52
_cerberus_That may be it then16:52
mtaylorhudson has all the build depends installed via apt-get from the nova-core ppa (although it's possible some things are installed from the soren-nova ppa as well)16:55
*** rlucio has joined #openstack16:56
*** rlucio has quit IRC17:12
*** rlucio has joined #openstack17:17
*** Ryan_Lane is now known as Ryan_Lane|away17:21
*** kevnfx has left #openstack17:21
*** scottie has quit IRC17:24
uvirtbotNew bug: #646177 in nova "DHCP server for flat network configuration" [Undecided,New]
*** scottie has joined #openstack17:31
edayvishy: around?17:44
*** rlucio has quit IRC17:55
*** jc_smith has joined #openstack17:58
*** maplebed has quit IRC18:01
*** pharkmillups has quit IRC18:04
jc_smithWith regard to Nova, I will have a flat network for each logical rack, so depending on what host is chosen for a vm to be instantiated on I need to allocate an IP and map it in DHCP according to the IP address range of that rack. I heard there is a more pluggable network branch coming soon, who would be best for me to talk to?18:05
PentheusMake your networks less flat and more bumpy.18:08
vishyeday: here18:21
edayvishy: I don't see a method for getting an instance by it's IP address anymore, this hidden someone or was it just removed?18:22
*** pvo has joined #openstack18:23
*** ChanServ sets mode: +v pvo18:23
*** rlucio has joined #openstack18:24
vishyeday: hmm18:27
*** rnirmal has quit IRC18:27
vishyeday:  instance_ref = db.fixed_ip_get_instance(None, address)18:28
*** perestrelka has quit IRC18:29
vishyeday: we need to clean up the naming of those functions: probably should be instance_get_by_address, eh?18:29
*** perestrelka has joined #openstack18:29
vishyeday: using your patch with the fix i posted, metadata works18:30
*** mtaylor has quit IRC18:30
jaypipesvishy, eday: got a sec to chat about the redis adapter?18:32
jaypipesvishy, eday: I'm running into an issue that I wanted your opinion on.  In SQLAlchemy, there are the wonderful Column classes that allow one to have differently typed (SQL types translated into Python types) columns.  In Redis, though, everything is a string.  I'm wondering whether to spend the time to develop the same Column class functionality or to force types to strings?  For instance, in the volume test case, we compa18:35
jaypipesre the result of volume_get(...).id with a number.  This works with SQLAlchemy, because the id attribute is a Column(Integer), but fails with Redis because it comes back as a string (as do all redis attributes)18:35
*** cloudmeat1 has quit IRC18:35
jaypipesvishy: so, the two options I see are a) coerce all attributes to strings (or use cast comparison with str(a) == str(b)) or b) create SQLAlchemy Column-like classes for the redis driver.18:36
*** sirp1 has joined #openstack18:40
edayjaypipes: hmm, seems we'd want them as their native type, but this is just what I was afriad of if we do - starting to rewrite bits of sqlalchemy for a shoe that doesn't really fit18:41
jaypipeseday: heh, I know.  but I've already re-written a good chunk ;)  I don't think this would be too difficult.18:42
edayvishy: cool, thanks. yeah, I looked for it in the 'instance_get*' functions and didn't see it there, probably should be18:42
*** pharkmillups has joined #openstack18:43
*** cloudmeat has joined #openstack18:52
*** krish_ has joined #openstack18:52
vishyjaypipes: i had an easy solution i was considering18:53
jaypipesvishy: yes?18:53
*** krish has quit IRC18:54
vishyif value[0].is_digit, convert to int18:54
vishythe only thing that leaves out is DateTime18:54
vishybecause we have no floats yet18:54
jaypipesvishy: yeah, that's a good solution.18:54
vishyguys, so we are seeing occasional greenthread related errors18:55
vishyRuntimeError: Second simultaneous read on fileno 9 detected.  Unless you really know what you're doing, make sure that only one greenthread can read any particular socket.  Consider using a pools.Pool. If you do know what you're doing and want to disable this error, call eventlet.debug.hub_multiple_reader_prevention(False)18:55
vishyI'm guessing is that this is related to AuthManager being a singleton18:56
edayvishy: is this with my branch?18:57
edayvishy: specifically, endpoint-cleanup?18:57
jaypipesvishy: I believe soren has a fix for that already in merge prop.18:57
vishyyes this is including yours and gundlachs branch18:58
jaypipesvishy: has to do with importing eventlet after daemonizing, so that the file descriptor epoll uses isn't screwed by the daemonize process.18:58
edayvishy: if so, I just fixed the queue consumer issue with that, pull my branch again18:58
*** blakeyeager has quit IRC18:59
vishyah ok18:59
vishybut i'm not running anything demonized18:59
jaypipesvishy: see soren's notes on that merge proposal...just rings a bell, that's all.18:59
vishyi'll pull newest versions from the branches and see if it helps19:00
jaypipesvishy: on a completely separate issue, is there a reason AOEManager.create_volume() doesn't ask the datastore to create the volume?  It assumes the datastore has already created the volume record...19:02
jaypipesvishy: which means that instead of only using the Manager class to manage volumes, the caller also has to directly call the datastore as well...19:02
*** rnirmal has joined #openstack19:04
*** jc_smith has quit IRC19:05
edayvishy: my last branch there fixes your problem exactly, I was just testing with concurrent requests before submitting for merge19:06
*** Ryan_Lane|away is now known as Ryan_Lane19:13
*** tonywolf has quit IRC19:15
vishyjaypipes: yes the creation is done in the api layer19:18
vishyjaypipes: round two of cleanup is19:18
vishyjaypipes: that code should be done in VolumeManager, and then the remote call happens to the VolumeService (running another volume manager on the other host)19:19
soren_cerberus_: I told you already.19:21
soren_cerberus_: I fixed it for you.19:21
soren09:03 <+soren> _cerberus_: Fixed your unit tests.19:21
soren09:03 <+soren> _cerberus_: I accept beer.19:21
soren09:08 <+soren> _cerberus_: lp:~soren/nova/rs_auth_fixes ( has the fix.19:21
soren09:09 <+soren> _cerberus_: To include it in your branch (not sure if you've done this before), just go into your checkout, do "bzr merge lp:~soren/nova/rs_auth_fixes; bzr  commit -m 'Merge fix from Soren'; bzr push"19:21
soren09:09 <+soren> _cerberus_: That's it.19:21
soren09:09 <+soren> _cerberus_: I ran the test suite on the hudson box with that patch applied and everything works brilliantly.19:21
vishygoing to lunch bbs19:22
soren_cerberus_: Over 10 hours ago. Try to keep up :)19:22
*** rlucio has quit IRC19:25
*** krish_ is now known as krish19:25
_cerberus_soren: I don't have my bouncer log this room when I'm away, so I must have missed it :-P19:26
soren_cerberus_: Well, you were online.19:26
_cerberus_Regardless, thank you19:26
*** krish_ has joined #openstack19:27
*** Cybodog has quit IRC19:27
*** krish_ has quit IRC19:28
*** maplebed has joined #openstack19:29
*** krish has quit IRC19:30
soren_cerberus_: Sure thing.19:35
jaypipesvishy: sorry, was on phone with eday... ok, that makes sense then.  did you want me to do that cleanup or did you already have that done somewhere?19:36
edaygundlach, vishy: Should we try putting our collective ec2-move branches back into trunk? Looks like things are working again19:38
_cerberus_Looks like my tests passed, and now I'm getting the same "deleting parent in endpoint" message that gundlach was receiving19:44
*** annegentle has left #openstack19:45
*** joearnold has joined #openstack19:48
*** joearnold has quit IRC19:55
*** joearnold has joined #openstack19:55
eday_cerberus_: yeah, I think we just need to clean out the .pyc files in there on hudson19:59
*** sophiap has quit IRC20:00
*** Pentheus has quit IRC20:00
*** f4m8_ has quit IRC20:00
*** rbergeron has quit IRC20:00
*** dendro-afk has quit IRC20:00
*** pjz has quit IRC20:00
*** loglaunch has quit IRC20:00
*** loglaunch has joined #openstack20:00
*** f4m8_ has joined #openstack20:00
*** dendro-afk has joined #openstack20:00
*** pjz has joined #openstack20:01
*** Pentheus has joined #openstack20:01
*** rbergeron has joined #openstack20:01
*** scottie has quit IRC20:05
soren_cerberus_, eday: I just did that and reapproved the merge. Let's se.20:07
sorensee, even.20:07
*** scottie has joined #openstack20:07
edayahh, so this will include a bunch of gundlach's branch... hm. we'll want to ge these other supporting branches in soon to fix trunk20:08
*** Podilarius has left #openstack20:09
edaygundlach: do you already have vish's metadata fix in your stuff somewhere?20:10
gundlachhi, back20:12
*** allsystemsarego has quit IRC20:12
gundlacheday: yes, let's put stuff into trunk.20:12
gundlacheday: which metadata fix, sorry?20:12
gundlacheday: if you mean the "i can't set a subdomain while testing locally" complaint in the mergeprop, then, yes20:13
_cerberus_soren: it merged. Thanks again20:13
soren_cerberus_: Hm. Ok. That's slightly unexpected, but cool.20:13
gundlachor do you mean the greenlet problem he mentioned?20:13
gundlach(scanning quickly thru the last few hours of logs)20:13
sorengundlach: Which greenlet problem is this?20:13
gundlach14:55 <vishy> guys, so we are seeing occasional greenthread related errors20:14
edaygundlach: the one in these comments:
soren_cerberus_: I tried the same thing yesterday for gundlach, and it didn't take.20:15
soren_cerberus_: (clearing out .pyc files, that is)20:15
*** jc_smith has joined #openstack20:15
soren_cerberus_: Hence my surprise. But great.20:15
edaysoren, gundlach: I fixed the shared fd problem20:15
_cerberus_That's odd20:15
gundlacheday: no, i haven't applied either of the diffs that are in your mergeprop comments20:15
_cerberus_Regardless, I'm happy.20:15
soreneday: Sorry, which one is that?20:16
gundlacheday: but, are you saying endpoint-cleanup is ready to merge into controllers-in-api?20:16
* soren is trying to catch up a bit20:16
edaygundlach: I got the cloupipe cleanup one, I didn't touch the MD one yet since that was in your changeset. I'll apply to my branch now though20:16
gundlacheday: all of my branches except tornado-excision have been applied to lp:~gundlach/nova/controllers-in-api, fyi20:16
edaygundlach: it will be in one sec, applying the vish's comment diff20:16
* gundlach is standing by to merge lp:~eday/nova/endpoint-cleaneup into controllers-in-api20:17
edaysoren: there was a concurrency issue in consumer fd's since it used to share a connection, I fixed in my endpoint-cleanup branch20:17
soreneday: /me looks20:17
soreneday: I'd like to know about the quirks of eventlet so that I can recognize them.20:18
soreneday: Hence my curiosity.20:18
*** Pentheus has quit IRC20:18
edaygundlach: commited, ready to merge, r31120:19
edaysoren: So, previously the rpc consumer for waiting for a response was a busy poll (every 100ms it would ask rabbitmq for a msg)20:20
edaysoren: I changed this to actually wait for the server to send, rather than a busy poll20:20
soreneday: Wicked.20:21
edaysoren: For now, it's one connection per queue response you're waiting on, but it was using the old code that shared a single connection for everything20:21
edaysoren: SO it was trying to use the same FD for reading multiple queue responses, which would break20:21
soreneday: I see, ok.20:21
edaysoren: so, not so much an eventlet thing, just a fix a busy-poll thing20:22
gundlachvishy: does your diff re: superuser role in rbac mean I no longer need to extend roles with FLAGS.superuser_roles?20:22
gundlacheday: would you mind pulling lp:~gundlach/nova/controllers-in-api now and see if there's anything that looks amiss?  i'm running the unit tests now, fwiw20:24
soreneday: Ah, ok.20:25
gundlacheday: so still has a tornado ioloop calling self.fetch every 100ms.  am i missing a branch?20:26
gundlach(in controllers-in-api after merging endpoint-cleanup)20:26
edaygundlach: gundlach that's not used anywhere now, I removed the thing that uses it. I couldn't remove the method since tests still used it20:26
gundlacheday: oh, that's great, ok20:27
edaygundlach: but your other branch seems to remove the tests using it, so we might be good to remove it all from rpc20:27
gundlacheday: yeah, though i'll do that after we get trunk fixed.20:27
edaygundlach: but we can do that after the controllers branch hits trunk20:27
edaygundlach: so, only other thing, and someone else piped up, is using subdomain for URL routing. I think this is the way to go for single server, but it does make testing a PITA.20:28
edaygundlach: So, we might want to just have a switch to default to rs or ec2, and an advanced feature of "support both on subdomain mapping"20:28
gundlacheday: been fixed, with FLAGS.FAKE_subdomain20:28
edaygundlach: oh, n/m then :)20:28
edaygundlach: I think we should bring soren's nova-api daemon branch in too...20:30
*** dendro-afk is now known as dendrobates20:32
*** dendrobates has joined #openstack20:32
*** ChanServ sets mode: +v dendrobates20:32
gundlacheday: merged nova-api-new-daemonize a few hours ago, is that the one you mean?20:33
edayoh, so it is. I must be screwing something up in my flags then20:34
edayoh, just --help is broken20:35
edaylooks ok here20:36
gundlachhmm, how come when i do a fresh pull of trunk, endpoint/ is gone?20:36
*** burris has quit IRC20:36
gundlachum, and nova.api.ec2 exists... looks like maybe r291 added it in?20:37
*** irahgel has left #openstack20:37
gundlachok, yes looks like r291 probably broke trunk again, as it merged all of the old trunk that this branch was based off of.  but i'm making a mergeprop now to put in controllers-in-api, so trunk won't stay broken long.20:38
edaygundlach: yeah, looks like _cerberus_ had it in his branch, probably branched from trunk when it was still in there20:38
gundlach_cerberus_: no worries, 10 minutes and we'll be back to normal :)20:39
eday_cerberus_: it's all good, about to go back in anyways ;)20:39
gundlachman eday you and i are on the same page for the last several minutes20:39
gundlacheday: kindly review
vishyjust caught up with scrollback20:41
vishygundlach: i was talking with people here and the consensus is that shortcutting is better and the superuser roles flag should go away completely20:42
gundlachvishy: not sure what "shortcutting is better" means?20:42
vishygundlach: the purpose of superuser is to bypass role checking completely, so having a role that bypasses other role checking is a little silly20:42
gundlachi think it means, i should integrate the diff you provided, and remove the append that i put in there?20:43
gundlachok, thanks!  will pull my change and add yours.20:43
gundlach(AFTER we merge to trunk ;) )20:43
vishycool, afterwards I'll submit a patch that removes FLAGS.superuser_roles20:43
*** joearnold has quit IRC20:45
*** tonywolf has joined #openstack20:47
sorenCan we all agree that removing a commit from trunk was a horrible idea, by the way?20:49
sorenThere's a revert operation. Use it.20:49
sorenDon't pretend shit never happened if you don't like it.20:49
vishysoren: yes, --overwrite was evil20:49
sorenI'm glad we're all on the same page. :)20:49
vishyeday: did i hear that you made call wait instead of poll?20:50
vishyeday: does it work in the twisted workers?20:50
edayvishy: I didn't touch the twisted workers, this is just for the API20:50
sorenThe twisted workers should be using txamqp anyways, really.20:51
sorenOr so it seems.20:51
edaysoren: yup, or *something* that doesn't poll :)  could do it with carrot+threads20:52
soreneday: That might work.20:52
sorenI'd be happy to work in this for Austin+1.20:52
sorencarrot+threads could be an experiment for me for tomorrow. Reworking things to work with txamqp is for Austin+1.20:54
edaysoren: at the summit we should discuss if twisted is needed for workers too, still seems like simple threads may do and would simplify a lot, but I don't want to get into that now :)20:54
soreneday: I'd like to take a stab at simplifying things while still using Twisted. There's a lot of potential for that.20:57
vishyeday: hmm, you didn't mess with
vishyeday: right now, is only used by api but there is some stuff it would be really nice to move into the workers21:00
edayvishy: I did rework rpc.call21:00
edayvishy: For the workers, we really just either need to do the twisted thing, or move off of twisted and just use carrot in a blocking fasion (wait)21:01
*** hornbeck_ has joined #openstack21:01
edayvishy: we can tackle that one at the summit :)21:02
*** littleidea has quit IRC21:02
*** hornbeck has quit IRC21:03
vishyeday: ok, so for now, it is a bad idea to try from a worker?21:05
*** kevnfx has joined #openstack21:05
edayvishy: yes.. and it was before since it was based on tornado ioloop21:05
vishyeday: also, did you merge termie's patch for removing tornado?  I think he pulled it out of the base test class and made it work better?21:05
edayvishy: nope, but gundlach may have21:06
*** hornbeck_ has quit IRC21:06
vishyeday: it did actually work though, because ioloop and twisted reactor somehow worked together21:06
edayvishy: oh, hmm. well we can make a for twisted that uses twisted primitives still21:07
gundlachi have not -- i started my own patch trying to remove BaseTestClass entirely but didn't get all the way done21:07
*** ctennis has quit IRC21:07
edayvishy: like most things, we're going to need the twisted and non-twisted version21:07
vishygundlach: take a look there, i think it fixes errors in some of the test cases that were slipping through as well.21:07
*** dendrobates is now known as dendro-afk21:07
gundlachvishy: link?21:08
edayvishy: and, actually, we could use as is in twisted if it's in it's own thread (defer thread)21:08
vishyeday: interesting idea21:08
gundlacheday: see anything wrong with controllers-in-api, or shall i go ahead and merge back into trunk?21:08
vishygundlach: trying an install with all the stuff merged in21:09
edaygundlach: I didn't, but I made a note for vishy to review :)21:09
edayvishy: ^^21:09
edaywant to make sure it's fixed for what they need21:09
gundlachvishy: sweet, thx, standing by21:09
vishygundlach: looks good21:12
gundlachhere we go21:12
*** jc_smith has quit IRC21:13
vishyi'll check all my branches and make sure they don't need to be updated21:13
*** jc_smith has joined #openstack21:13
vishyi'd like to get that stuff in soon as well as termie's ioloop removal.  Can we finally kill tornado????21:14
gundlachvishy: merging to trunk now.  assuming it works, termie already has an Approve from jmck, so he should be able to merge it.21:15
*** mtaylor has joined #openstack21:15
*** ChanServ sets mode: +v mtaylor21:15
mtaylor_cerberus_: any idea what the problem wound up being?21:15
*** burris has joined #openstack21:16
gundlachok, controllers-in-api merged.21:17
gundlacheday, soren, vishy, thanks for your reviews, code, input, and patience.21:18
gundlachand i am now out for the week.  see you monday.  call 678 439 6684 if i'm blocking anybody else!21:18
_cerberus_mtaylor: soren fixed it with one of his branches. Apparently some kind of incorrect usage in a conditional from something I pulled in from trunk21:19
*** sophiap has joined #openstack21:19
_cerberus_mtaylor: plus soren cleared all of the pyc files out of nova/endpoint21:19
mtaylorsoren is good at fixing thngs :)21:20
_cerberus_That's something, anyway ;-)21:20
*** kevnfx has quit IRC21:20
*** ctennis has joined #openstack21:20
*** gundlach has quit IRC21:21
vishyI'm good at breaking things!21:21
*** rlucio has joined #openstack21:22
*** scottie has quit IRC21:22
sorenmtaylor: I owe it all to being stubborn.21:26
*** Pentheus has joined #openstack21:27
*** annegentle has joined #openstack21:29
*** Cybodog has joined #openstack21:36
*** scottie has joined #openstack21:38
*** kevnfx has joined #openstack21:45
*** kevnfx has quit IRC21:53
*** joearnold has joined #openstack21:56
*** scottie has quit IRC21:59
*** jc_smith has quit IRC22:04
*** Podilarius has joined #openstack22:06
*** pvo has quit IRC22:07
*** sirp1 has quit IRC22:10
*** metcalfc has quit IRC22:16
*** metcalfc has joined #openstack22:17
*** annegentle has left #openstack22:20
*** jdarcy has quit IRC22:23
*** johnpur has quit IRC22:25
*** krzycoder has joined #openstack22:27
*** littleidea has joined #openstack22:33
*** rnirmal has quit IRC22:35
*** joearnold has quit IRC22:42
*** tonywolf has quit IRC22:42
*** rlucio_ has joined #openstack22:46
*** rlucio has quit IRC22:49
*** rlucio_ is now known as rlucio22:49
*** joearnold has joined #openstack22:50
*** zheng_li has quit IRC22:54
*** littleidea_ has joined #openstack22:57
*** littleidea has quit IRC22:57
*** littleidea_ is now known as littleidea22:57
*** miclorb_ has joined #openstack23:01
*** klord has quit IRC23:13
*** pharkmillups has quit IRC23:14
*** dizz has joined #openstack23:30
*** dizz|away has quit IRC23:32
*** Cybodog has quit IRC23:33
*** rlucio has quit IRC23:35
*** maple_bed has joined #openstack23:44
*** maplebed has quit IRC23:45
*** Pentheus has quit IRC23:49
*** maplebed has joined #openstack23:50
*** howardroark has joined #openstack23:52
*** abecc has quit IRC23:53
*** maplebed has quit IRC23:54

Generated by 2.14.0 by Marius Gedminas - find it at!