Wednesday, 2014-02-12

« Tuesday, 2014-02-11 Index Thursday, 2014-02-13 »
*** chandan_kumar has joined #openstack-climate03:51
*** DinaBelova_ is now known as DinaBelova05:45
*** saju_m has joined #openstack-climate06:03
*** YorikSar has quit IRC08:06
*** YorikSar has joined #openstack-climate08:06
*** YorikSar has quit IRC08:07
*** YorikSar has joined #openstack-climate08:07
*** YorikSar has quit IRC08:08
*** YorikSar has joined #openstack-climate08:08
*** bauzas has joined #openstack-climate08:12
openstackgerritA change was merged to stackforge/climate: Modifying coverage tox.ini configuration  https://review.openstack.org/7267408:46
openstackgerritA change was merged to stackforge/climate: Remove old Scheduler RPC Service  https://review.openstack.org/7261208:47
*** DinaBelova is now known as DinaBelova_09:40
*** saju_m has quit IRC10:57
*** YorikSar has quit IRC11:14
*** YorikSar has joined #openstack-climate11:24
*** YorikSar has quit IRC11:27
*** YorikSar has joined #openstack-climate11:30
*** YorikSar has quit IRC11:31
*** YorikSar has joined #openstack-climate11:32
*** YorikSar has quit IRC11:33
*** YorikSar has joined #openstack-climate11:34
*** YorikSar has quit IRC11:34
*** YorikSar has joined #openstack-climate11:34
*** YorikSar has quit IRC11:35
*** YorikSar has joined #openstack-climate11:35
*** saju_m has joined #openstack-climate11:40
*** YorikSar has quit IRC11:48
*** saju_m has quit IRC11:49
*** YorikSar has joined #openstack-climate11:51
*** YorikSar has quit IRC11:51
*** YorikSar has joined #openstack-climate11:52
*** DinaBelova_ is now known as DinaBelova12:04
*** YorikSar has quit IRC12:32
*** YorikSar has joined #openstack-climate12:33
*** casanch1 has joined #openstack-climate13:03
*** saju_m has joined #openstack-climate13:05
*** saju_m has quit IRC13:06
*** saju_m has joined #openstack-climate13:07
openstackgerritChristian Martinez proposed a change to stackforge/climate: Adding tests for sqlalchemy class  https://review.openstack.org/7294313:34
*** pafuent has joined #openstack-climate13:36
openstackgerritNikolay Starodubtsev proposed a change to stackforge/python-climateclient: Add oslo modules for tests  https://review.openstack.org/7294513:40
openstackgerritNikolay Starodubtsev proposed a change to stackforge/python-climateclient: Add test framework for climateclient  https://review.openstack.org/7229613:51
openstackgerritChristian Martinez proposed a change to stackforge/climate: Adding tests for sqlalchemy class  https://review.openstack.org/7294313:53
*** cmart has joined #openstack-climate13:55
*** YorikSar_ has joined #openstack-climate13:58
*** cmart has quit IRC14:03
*** cmart_ has joined #openstack-climate14:03
*** YorikSar has quit IRC14:03
*** YorikSar_ is now known as YorikSar14:03
*** saju_m has quit IRC14:21
DinaBelovaguys, small thing I'd like to post here14:24
DinaBelovahttps://etherpad.openstack.org/p/climate-incubation-status14:25
DinaBelovaIt's our current incubation status with thing we should do to go further14:25
DinaBelovathere are links there to started incubation application14:25
DinaBelovaand things that should be fixed14:25
DinaBelovathere are undone issues in bold14:26
DinaBelovabauzas, Nikolay_St, swann, pafuent, casanch1, f_rossigneux ^^^^^^14:27
bauzasI'm commenting it now14:27
casanch1ok, I'll take a look14:28
pafuentHi. I'll take a look too14:29
DinaBelovathanks, guys14:29
pafuentI added the links to the devstack reviews14:37
*** saju_m has joined #openstack-climate14:38
openstackgerritChristian Martinez proposed a change to stackforge/climate: Adding tests for sqlalchemy class  https://review.openstack.org/7294314:52
*** cmart_ has left #openstack-climate14:55
*** cmart__ has joined #openstack-climate14:56
*** YorikSar_ has joined #openstack-climate14:59
*** YorikSar has quit IRC15:00
*** YorikSar_ is now known as YorikSar15:01
*** chandan_kumar has quit IRC15:39
*** chandan_kumar has joined #openstack-climate15:40
*** bauzas has quit IRC15:56
*** chandan_kumar has quit IRC15:59
*** chandankumar_ has joined #openstack-climate15:59
casanch1Hi, I'm updating oslo in climate16:00
casanch1using the Dina's patchset, which removed rpc and notifier16:01
casanch1but the oslo-incubator update.py script still updates rpc and notifier16:01
casanch1http://paste.openstack.org/show/64674/16:01
casanch1do you know what could be wrong?16:01
DinaBelovamoment16:02
casanch1I'll be right back16:02
DinaBelovaI have one guess, but I thought I fixed it in my change16:02
DinaBelovayep, I fixed it - I moved notifier and rpc modules from openstack-common.conf16:03
DinaBelovahttps://review.openstack.org/#/c/57880/13/openstack-common.conf16:03
DinaBelovathat's strange it's trying still to update them :(16:03
*** bauzas has joined #openstack-climate16:05
DinaBelovaI only can imagine that somehow there is set rpc as dependency for https://github.com/openstack/oslo-incubator/blob/master/openstack/common/service.py#L47 - but as I see how update.py works it should not - https://github.com/openstack/oslo-incubator/blob/master/update.py#L24316:08
DinaBelovaI'll try some greps, moment16:14
DinaBelovaI'm quite surprised, really16:15
DinaBelovait should not do that16:16
DinaBelovaso there are two variants: 1/ u may update it and remove rpc and notifies manually16:18
DinaBelova2/ u may ask oslo-incubator guys if they see smth strange there16:18
casanch1ok, I will remove rpc and notify and run the tests in climate16:36
DinaBelovaok16:39
DinaBelovabtw 1/ + 2/ is the best variant :)16:39
*** bauzas has quit IRC16:53
*** cmart__ has quit IRC17:09
*** casanch1 has quit IRC17:16
*** casanch1 has joined #openstack-climate17:18
*** bauzas has joined #openstack-climate17:29
*** bauzas has quit IRC17:36
*** bauzas has joined #openstack-climate17:38
openstackgerritSylvain Bauza proposed a change to stackforge/climate: Port to Pecan/WSME for API v2  https://review.openstack.org/7101117:46
*** saju_m has quit IRC17:50
*** saju_m has joined #openstack-climate17:51
*** saju_m has quit IRC17:58
*** saju_m has joined #openstack-climate18:00
*** chandankumar_ has quit IRC18:03
*** bauzas has quit IRC18:07
pafuentDinaBelova: Are you around?18:08
SergeyLukjanovpafuent, hey, looks like your devstack patch should work ok now18:10
SergeyLukjanovat least w/o clone error ;)18:10
DinaBelovapafuent, sometimes18:10
pafuentSergeyLukjanov: I hope18:10
pafuentSergeyLukjanov: I'm running a recheck no bug to test it18:11
SergeyLukjanovpafuent, yup, see it18:11
pafuentDinaBelova: There is a reason to get all events in ManagerService._event() and then only process the first one?18:12
DinaBelovait's because we're running events that should happen18:13
DinaBelovaspeaking about time18:13
DinaBelovaso they are got from DB sorted by date18:13
DinaBelovaand the first one is the earliest18:13
pafuentpafuent: IMO that is a memory/performance issue.18:13
DinaBelovayes, it is18:14
pafuentpafuent: I could fill a bug with that18:14
DinaBelovabut we're going to rewrite it with HA with multiple events workers18:14
DinaBelovalike there are lots of workers and they'll use other type of behaviour18:14
DinaBelovaI dunno if that should be fixed in current code status - if it'll be changed anyway. Unless you're speaking about some other type of optimisation18:15
DinaBelovahow do you suggest to fix that?18:16
pafuentDinaBelova: A really simple aproach, avoid the use of query.all() and use query.first()18:16
DinaBelovapafuent, oh, ok :) That's nice (but keep in mind sorting thing)18:17
pafuentDinaBelova: The idea is to use the same db_api method but returning only the first event18:17
DinaBelovaoh, gotcha18:17
DinaBelovaI'm ok with that18:17
DinaBelovaalthough I don't think it will help much with performance :D18:18
pafuentDinaBelova: This will avoid a lot of Event objects in memory and the access to the DB will be reduced18:18
DinaBelovaok, nice18:18
pafuentDinaBelova: So, Do you think that will be useful to fill a bug and fix that?18:19
DinaBelovaI think that's nothing bad with it :)18:20
pafuentDinaBelova: Ok, I'll fill the bug18:20
*** saju_m has quit IRC18:51
*** casanch1 has quit IRC18:54
*** saju_m has joined #openstack-climate19:01
*** casanch1 has joined #openstack-climate19:17
*** bauzas has joined #openstack-climate19:33
openstackgerritDina Belova proposed a change to stackforge/climate: Port to oslo.messaging  https://review.openstack.org/5788019:39
*** saju_m has quit IRC19:48
pafuentSergeyLukjanov: What should I do in order to scp the screen logs for climate?19:52
SergeyLukjanovpafuent, it should be uploaded automatically19:57
pafuentSergeyLukjanov: So, if no logs were uploaded, that implies that the screens failed to start?20:01
SergeyLukjanovpafuent, I think yes20:08
pafuentSergeyLukjanov: Ok, I'll try to find what happened20:10
pafuentDinaBelova: Regarding to https://bugs.launchpad.net/climate/+bug/127945920:11
pafuentDinaBelova: I found that the something similar happens in update_lease20:11
pafuentDinaBelova: Should I fix that in the same patch or fill a new bug for that?20:12
bauzasDinaBelova: that's planned to be rewritten in a bp20:12
bauzasoops20:12
bauzaspafuent: instead20:12
DinaBelovanah, one moment20:13
bauzaspafuent: the current _event() consumer is a PC20:13
bauzasPOC20:13
DinaBelovaI'm reading logs20:13
DinaBelovabauzas, yes, we've discussed that20:13
DinaBelovabut pafuent sujjest to remove at least some moments like20:13
bauzasDinaBelova: I was speaking to pafuent20:13
bauzas:)20:13
DinaBelovausing all instead of first20:13
bauzaswell, that's something we spotted a while ago20:13
pafuentAhhh20:14
bauzasand the main point is that if we do that, that could lead to potential lock-in20:14
pafuentThe fix is really simple.20:14
pafuentbauzas: Why?20:14
bauzasthe ideal way would be to consume all tasks in a separate distributed service20:14
DinaBelova[22:14:27]  <DinaBelova> but we're going to rewrite it with HA with multiple events workers20:15
DinaBelova[22:14:54]  <DinaBelova> like there are lots of workers and they'll use other type of behaviour20:15
DinaBelova[22:15:44]  <DinaBelova> I dunno if that should be fixed in current code status - if it'll be changed anyway. Unless you're speaking about some other type of optimisation20:15
DinaBelova[22:16:03]  <DinaBelova> how do you suggest to fix that?20:15
DinaBelova[22:16:32]  <pafuent> DinaBelova: A really simple aproach, avoid the use of query.all() and use query.first()20:15
DinaBelovathat was our conversation here some time ago20:15
bauzaswell, that's risky because that's not transactional20:15
bauzasthe current thread will take the DB lock20:16
DinaBelovabauzas, you mean using first instead of all?20:16
DinaBelovais it non-transactional?20:16
bauzasah, didn't catched the .first() thing20:16
DinaBelovathat's really small change :)20:17
pafuentThe idea is to reduce the memory used and improve the performance a little20:17
bauzaswell, I was speaking of processing all the undone events all in the same thrzad20:17
bauzasbut yes, indeed20:17
DinaBelova:)20:17
pafuentBTW, I found that the same is done in update_lease20:17
bauzasthat's a small change, +1 for .first()20:17
DinaBelovathat's difficult topic to understand :S bauzas $ I immediately go to discussion of HA workers :D20:18
DinaBelovapafuent, if there is smth like that in update_lease - please rewrite bug description and fix that too in one change20:18
DinaBelovathat won't be much20:18
bauzasDinaBelova: +120:18
pafuentbauzas: DinaBelova: Ok, I'll do it20:18
DinaBelova:D20:19
pafuentbauzas: DinaBelova: Thanks20:19
DinaBelovapafuent, sorry for misunderstanding one more time :D20:19
pafuentDinaBelova: No problem20:19
bauzasthat's pretty late for both of us ;)20:19
DinaBelovacause HA architecture is in our minds and we can see that everywhere :D20:19
DinaBelovaand yep, it's late :)20:19
DinaBelovabauzas, u've got utc+2 or utc+3 now?20:20
DinaBelovaor utc+1?20:20
bauzasutc+120:20
DinaBelovaok, gotcha20:20
bauzasutc+2 in summer20:20
bauzas2130 now20:20
DinaBelovain Russia we now have no summer/winter time20:20
bauzasDinaBelova: I know :)20:20
DinaBelovaand that's uncomfortable :(20:20
bauzasand I'll particularly appreciate our friday meetings on summer time...20:21
bauzasthat's joke :)20:21
bauzas1700-180020:21
DinaBelova:D20:21
bauzasI'll appreciate if we'll be quick :)20:21
DinaBelovaI'm guess it will be much better to use permanent utc+3 instead of utc+4 if that's need not to change it in my country20:21
DinaBelovaI guess ***20:21
bauzaswell, i do definitely prefer summer itme20:21
bauzastime20:21
DinaBelova:)20:21
bauzaswell, this HA thing is braining me out20:22
bauzasI'm thinking of the possibilities and what's done with other projects20:22
DinaBelovain my place that permanent utc+4 leads to lack of sun for fall/winter/spring20:22
DinaBelova:(20:22
bauzasaiui, Taskflow could be some help20:22
DinaBelovanot sure :? Taskflow cannot use distributed flows now, as I remember20:23
bauzasthat's Mistral job20:23
DinaBelovayep20:23
DinaBelovaanyway, I20:23
DinaBelovaam going to find some solution for that20:23
bauzasI would love be part of the thinkings ,)20:24
bauzasI'm currently focusing on multi-threading concerns in Python20:24
bauzasfor my knowledge20:24
bauzasand that's kind of work I would love having20:24
bauzasprovided we need to keep a stateless architecture20:25
DinaBelovabauzas, sure20:25
DinaBelovaI guess YorikSar will help us here :)20:25
DinaBelovahe loves such tasks :)20:25
bauzas:)20:26
bauzaswell, the few things I can see with multi-processing is on the RPC fanout way20:27
bauzasthat's not what we need20:27
bauzasmaybe a climate-conductor would be necessary before crafting HA20:27
bauzasand I definitely think we need objects for the Manager20:28
*** DinaBelova is now known as DinaBelova_20:47
*** casanch1 has quit IRC21:02
*** pafuent has left #openstack-climate21:13
openstackgerritSylvain Bauza proposed a change to stackforge/climate: Port to Pecan/WSME for API v2  https://review.openstack.org/7101123:26
*** bauzas has quit IRC23:35
« Tuesday, 2014-02-11 Index Thursday, 2014-02-13 »

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!