Wednesday, 2014-01-22

« Tuesday, 2014-01-21 Index Thursday, 2014-01-23 »
*** flwang has quit IRC00:02
*** amitgandhi has joined #openstack-marconi00:08
*** amitgandhi has quit IRC00:12
*** flwang has joined #openstack-marconi00:20
*** balajiiyer has joined #openstack-marconi00:21
*** oz_akan_ has joined #openstack-marconi00:24
*** reed has quit IRC00:28
*** rongze has joined #openstack-marconi00:35
*** rongze has quit IRC00:40
*** flaper87 is now known as flaper87|afk00:42
*** reed has joined #openstack-marconi00:44
*** reed has quit IRC00:48
*** kgriffs_afk is now known as kgriffs00:50
*** amitgandhi has joined #openstack-marconi01:09
*** kgriffs is now known as kgriffs_afk01:10
*** amitgandhi has quit IRC01:15
*** nosnos has joined #openstack-marconi01:35
*** malini_afk is now known as malini01:43
*** rongze has joined #openstack-marconi01:46
*** kgriffs_afk is now known as kgriffs01:49
*** renlt has joined #openstack-marconi02:05
*** haomaiwa_ has quit IRC02:10
*** haomaiwang has joined #openstack-marconi02:10
*** haomaiwa_ has joined #openstack-marconi02:15
*** haomaiwang has quit IRC02:17
*** malini is now known as malini_afk02:51
*** amitgandhi has joined #openstack-marconi03:12
*** amitgandhi has quit IRC03:17
*** kgriffs is now known as kgriffs_afk04:05
*** amitgandhi has joined #openstack-marconi04:13
*** amitgandhi has quit IRC04:17
*** liang has joined #openstack-marconi04:22
*** balajiiyer has quit IRC04:31
*** rongze has quit IRC04:52
*** chandankumar_ has joined #openstack-marconi05:08
openstackgerritA change was merged to openstack/marconi: fix: ConnectionError no longer takes positional arguments  https://review.openstack.org/6816105:11
*** amitgandhi has joined #openstack-marconi05:13
*** amitgandhi has quit IRC05:18
*** rongze has joined #openstack-marconi05:23
*** rongze has quit IRC05:28
*** rongze has joined #openstack-marconi05:35
*** flwang has quit IRC05:36
*** rongze has quit IRC05:40
*** flwang has joined #openstack-marconi05:59
*** oz_akan_ has quit IRC06:11
*** amitgandhi has joined #openstack-marconi06:14
*** rongze has joined #openstack-marconi06:17
*** amitgandhi has quit IRC06:18
*** oz_akan_ has joined #openstack-marconi06:42
*** nosnos_ has joined #openstack-marconi06:43
*** nosnos has quit IRC06:43
*** oz_akan_ has quit IRC06:47
*** amitgandhi has joined #openstack-marconi07:15
*** amitgandhi has quit IRC07:19
*** oz_akan_ has joined #openstack-marconi07:43
*** oz_akan_ has quit IRC07:48
*** flaper87|afk is now known as flaper8708:03
*** amitgandhi has joined #openstack-marconi08:15
*** renlt has quit IRC08:18
*** amitgandhi has quit IRC08:20
openstackgerritFlavio Percoco proposed a change to openstack/marconi: Sync oslo-cache from oslo-incubator  https://review.openstack.org/6812108:36
*** oz_akan_ has joined #openstack-marconi08:44
*** oz_akan_ has quit IRC08:48
*** ykaplan has joined #openstack-marconi09:10
*** amitgandhi has joined #openstack-marconi09:16
liangflaper87, there?09:17
flaper87liang: yes sir09:17
liangflaper87, ;)09:18
liangI have a question about the usage of the client lib - I always need to ensure the endpoint url contains '/v1' at the end. I am not sure if I missed something or it's supposed to be used that way. I proposed a patch for devstack on https://review.openstack.org/#/c/60754/ a while ago, but the feedback was that the version info would be eliminated even from the other APIs.09:18
flaper87liang: mhh, that's a good point. I didn't put it in keystone on purpose09:20
flaper87because of what Dean said09:21
*** amitgandhi has quit IRC09:21
flaper87but I'm not sure there's actually a good way to append this to the url returned by keystone. I don't recall adding it09:21
flaper87mmhh09:21
flaper87let me check09:21
liangI think keystone will just store the endpoint untouched.09:22
liangIn that case, we cannot directly consume the endpoint retrieved from Keystone.09:23
flaper87oh no, I now remember09:23
flaper87I did add it09:23
flaper87https://github.com/openstack/python-marconiclient/blob/master/marconiclient/transport/http.py#L6009:23
flaper87marconiclient should add the version lable to the endpoint09:23
flaper87or better, the http transport should09:23
flaper87liang: if you needed that, it means that it's not working09:24
flaper87mmhh09:24
lianghehe09:24
flaper87liang: can you debug that line?09:25
flaper87I don't have an environment with keystone and marconi installed handy09:25
flaper87I'll setup one, though09:25
liangI seems I have to give the full endpoint with the latest client.09:25
* flaper87 does that09:25
liangok09:25
liangI will take a look09:25
flaper87liang: how are you creating the client instance ?09:25
liangflaper87,  http://paste.openstack.org/show/61665/09:26
liangThat was extracted from https://review.openstack.org/#/c/61036/6/contrib/marconi-plugin/plugin/queue.py09:27
flaper87mmh, you know marconi can do that for you, don't you?09:29
flaper87I mean, getting the remote url09:29
flaper87from keystone09:30
flaper87liang: https://github.com/openstack/python-marconiclient/blob/master/marconiclient/queues/v1/client.py#L4509:30
flaper87just pass the 'auth_opts' dict in the conf dict to the client instance09:30
flaper87oh well, you're basically doing that09:31
flaper87for auth09:31
flaper87mmh, probably that linde is a bit dangerous09:33
liangyeah, the endpoint passed to client is also retrieved from keystone. And I have tested that09:33
flaper87liang: https://github.com/openstack/python-marconiclient/blob/master/examples/simple.py this works09:36
flaper87the only difference is that the example doesn't use auth_opts09:36
flaper87liang: I'll have to dig more into that :( let me setup an environement with keysotne and marconi09:39
*** oz_akan_ has joined #openstack-marconi09:44
*** flwang has quit IRC09:47
*** oz_akan_ has quit IRC09:49
liangflaper87,  Thanks!09:49
liangflaper87, a quick question, was the intention of https://github.com/openstack/python-marconiclient/blob/master/marconiclient/transport/http.py#L60 to add the /v1 prefix to the str ref?09:50
flaper87liang: that line removes the version from the reference because href are not suppose to have the version number, it should be set by the API itself09:54
flaper87I now have some doubts about that line09:54
flaper87also, there seems to be a missing append of the API version to the URL somewhere there09:54
flaper87:/09:55
liangThanks! It seems the version number should be striped off from request.endpoint if it is there. Then add the version prefix to ref. So later when endpoint and ref is concatenated together the version number will be there.10:01
flaper87liang: yeah, more like "prepend the API version if it's not already there"10:02
liangflaper87, yeah10:03
liangflaper87,  BTW, regarding the heat plugin patch, I moved the code to Heat contrib directory to speed up the review process hopefully.10:05
flaper87liang: cool, it makes sense. I guess you can move out of there when marconi will be integrated10:06
liangflaper87, yeah, exactly what I planed to do ;)10:06
flaper87awesome :D10:06
liang;)10:06
*** amitgandhi has joined #openstack-marconi10:17
*** amitgandhi has quit IRC10:21
*** flwang has joined #openstack-marconi10:31
*** liang has quit IRC10:39
*** oz_akan_ has joined #openstack-marconi10:45
*** oz_akan_ has quit IRC10:49
*** ykaplan has quit IRC11:00
openstackgerritFlavio Percoco proposed a change to openstack/marconi: DRY applied to class names  https://review.openstack.org/6835211:09
*** amitgandhi has joined #openstack-marconi11:17
*** andreykurilin has joined #openstack-marconi11:18
*** amitgandhi has quit IRC11:22
*** rongze has quit IRC11:26
*** rongze has joined #openstack-marconi11:27
*** rongze has quit IRC11:31
openstackgerritAndrey Kurilin proposed a change to openstack/python-marconiclient: Fix wrong client name in utils  https://review.openstack.org/6835611:41
*** oz_akan_ has joined #openstack-marconi11:46
*** oz_akan__ has joined #openstack-marconi11:47
*** oz_akan_ has quit IRC11:51
*** oz_akan__ has quit IRC11:52
*** rongze has joined #openstack-marconi12:09
*** rongze has quit IRC12:14
*** amitgandhi has joined #openstack-marconi12:18
*** amitgandhi has quit IRC12:22
*** rongze has joined #openstack-marconi12:39
openstackgerritAndrey Kurilin proposed a change to openstack/python-marconiclient: Delete unused module  https://review.openstack.org/6835612:39
openstackgerritFlavio Percoco proposed a change to openstack/marconi: Don't use `override_time` but mock instead  https://review.openstack.org/6836512:40
*** tedross has joined #openstack-marconi12:42
*** mpanetta has joined #openstack-marconi12:55
*** ykaplan has joined #openstack-marconi13:14
*** amitgandhi has joined #openstack-marconi13:19
*** vkmc has joined #openstack-marconi13:23
*** vkmc has quit IRC13:23
*** vkmc has joined #openstack-marconi13:23
*** amitgandhi has quit IRC13:23
*** rongze has quit IRC13:37
*** thomasem has joined #openstack-marconi13:43
*** rongze has joined #openstack-marconi13:47
*** cpallares has joined #openstack-marconi13:56
*** oz_akan_ has joined #openstack-marconi14:06
*** rongze_ has joined #openstack-marconi14:14
*** rongze has quit IRC14:16
*** oz_akan_ has quit IRC14:17
*** haomaiwang has joined #openstack-marconi14:17
*** rongze has joined #openstack-marconi14:17
*** oz_akan_ has joined #openstack-marconi14:18
*** rongze_ has quit IRC14:19
*** haomaiwa_ has quit IRC14:19
*** jcru has joined #openstack-marconi14:20
*** rongze_ has joined #openstack-marconi14:22
*** rongze has quit IRC14:25
*** amitgandhi has joined #openstack-marconi14:33
*** amitgandhi has quit IRC14:33
*** ykaplan has quit IRC14:34
*** amitgandhi has joined #openstack-marconi14:34
*** ykaplan has joined #openstack-marconi14:37
*** rongze_ has quit IRC14:49
*** balajiiyer has joined #openstack-marconi14:51
*** nosnos_ has quit IRC14:57
*** ametts has quit IRC14:58
*** zyuan has joined #openstack-marconi14:58
*** rongze has joined #openstack-marconi15:05
*** amitgandhi has quit IRC15:07
*** rongze has quit IRC15:22
*** malini_afk is now known as malini15:22
balajiiyerMorning15:23
balajiiyerkgriffs: Has the tarball been cut ? :)15:23
flaper87balajiiyer: morning :)15:23
flaper87balajiiyer: http://tarballs.openstack.org/marconi/15:23
flaper87nope15:23
flaper87:)15:23
*** amitgandhi has joined #openstack-marconi15:25
balajiiyerAh. ok. Zuul seems to be sane. so thats good news.15:26
*** amit has joined #openstack-marconi15:26
*** amit is now known as Guest7786215:27
flaper87balajiiyer: define sane, pls15:27
flaper87:D15:27
*** amitgandhi has quit IRC15:29
*** kgriffs_afk is now known as kgriffs15:30
kgriffso/15:31
* kgriffs hates traffic15:31
flaper87kgriffs: hey hey!15:31
balajiiyeroh, when I looked at it this morning queued events were down to 0.  and not 1900+. :)15:31
kgriffsheh15:31
* flaper87 hates traffic too15:31
kgriffsheck of a commute today - I had to drive from Austin to San Antonio. :p15:31
balajiiyerkgriffs: I thought you liked traffic. Cars *queued* up on roads.15:32
kgriffsLOL15:32
kgriffsso, I'm not seeing a branch and/or tag on openstack/marconi15:33
kgriffsand no icehouse-2 tarball15:34
kgriffsso...15:34
kgriffsnot sure what is happening there15:34
kgriffswas supposed to be done by now15:34
flaper87kgriffs: the cut hasn't been done, AFAIK15:34
kgriffsflaper87: looks like it15:34
kgriffsmaybe ttx slept in. :p15:34
kgriffsflaper87: looks like ttx is making the rounds now.15:35
kgriffsmalini: ping15:36
malinikgriffs: pong15:36
flaper87kgriffs: just saw your message15:39
kgriffshey, so i benchmarked that patch15:39
flaper87kgriffs: what patch is pending?15:39
kgriffsthe atomic one15:39
kgriffsi'm not sure it should go in...15:39
kgriffswanted to discuss it real quick15:39
flaper87I thought we said no15:39
flaper87sure15:39
malinikgriffs: steppiong away for a meeting..will be back in an hr15:39
kgriffsso, i think it is ok to slip it. any objections?15:39
kgriffsmalini: ok... real quick, the servers you gave me were a bit misconfigured15:40
kgriffsreally old version of marconi15:40
kgriffswrong ip addressed, etc.15:40
malinikgriffs: hmmmm..tht is embarassing15:40
flaper87I was fine with it, TBH15:41
kgriffsno problem, I just wanted to share what i did to update them15:41
kgriffsflaper87: ok15:41
maliniyou want to update it?15:41
flaper87I even +2'd :D15:41
malinikgriffs: ^15:41
kgriffsmalini: let's chat more after your meeting15:41
maliniok15:41
flaper87that based on the thoughts that we'll need it and that we can improve it during i-315:41
kgriffsbut, my benchmarks showed no difference after the patch15:41
kgriffsflaper87: ah, makes sense15:41
kgriffswell...15:41
kgriffsi think we might as well do it then. Code looks good, and benchmarks looks good - we can do more rigorous testing in i-315:42
kgriffssound OK?15:42
flaper87kgriffs: lets go for it15:42
kgriffslooks like alej +2d so u can approve15:42
flaper87+A15:43
kgriffskk15:43
kgriffscome on, zuul!15:47
kgriffsany moment now...15:47
kgriffsman15:51
kgriffssoooo slow15:52
flaper87hehehe15:57
flaper87patience, young padawan15:58
*** chandankumar_ has quit IRC15:58
openstackgerritA change was merged to openstack/marconi: fix(mongodb): Ensure batch message posts are atomic (all or nothing)  https://review.openstack.org/6732015:59
flaper87w00000000000000000000000000000000t15:59
*** Guest77862 has quit IRC16:01
*** chandankumar_ has joined #openstack-marconi16:01
*** sballe has joined #openstack-marconi16:02
*** jamieh has joined #openstack-marconi16:02
*** amitgandhi has joined #openstack-marconi16:16
*** amit has joined #openstack-marconi16:17
*** amit is now known as Guest476616:17
*** Guest4766 is now known as amitgandhi_16:17
*** amitgandhi_ has quit IRC16:18
*** amitgandhi has quit IRC16:20
*** amitgandhi has joined #openstack-marconi16:21
*** tedross has quit IRC16:33
malinikgriffs: sorry abt the whole mess up16:35
maliniWas trying to do too many things at once16:35
maliniBut we really need to get the benchmarking stuff in our jenkins pipeline16:36
maliniI'll start working on it some time after mid Feb16:37
kgriffsoz_akan_: some other dbs I have started looking at that could be alternatives to mongodb - http://www.orientechnologies.com/orientdb-vs-mongodb/, https://foundationdb.com, http://voltdb.com/16:41
kgriffsmalini: hey, no worries.16:41
kgriffs+1 for getting it into our pipeline16:41
kgriffsanybody heard of a sort of "travis" for load testing?16:42
kgriffs(Travis CI)16:42
* flaper87 doesn't recommend Orient. Not yet!16:43
oz_akan_kgriffs: voltdb is too radical16:43
oz_akan_if it works it might work well16:43
oz_akan_I am not aware of others16:44
oz_akan_we need a db which can shard automatically16:44
oz_akan_or use ring (consistent hashing)16:44
oz_akan_I will reboot my laptop16:44
*** oz_akan_ has quit IRC16:44
kgriffsvoltdb is backed by some hard-core research - not really "radical" as in "risky" i would say, but "radical" as in NewSQL16:45
*** tedross has joined #openstack-marconi16:45
*** ykaplan has quit IRC16:45
kgriffsoz_akan_: mongodb can shard automatically16:45
kgriffsflaper87: is orient too new or what?16:46
flaper87kgriffs: yeah, sorry, that was to vague16:46
flaper87so, it's not too bad. It's not too new either. IMHO, it's too many things. Also, I've close friends working with it and although they like it, the don't think it's there yet16:47
flaper87the comment is still vague but that's what I've got. It's all based on references16:47
flaper87from other trusted devs16:47
flaper87speaking of rings: https://github.com/flaper87/rust-ring16:48
flaper87:D16:48
flaper87I still have to complete it, I'll put some time on it today16:48
*** whenry has joined #openstack-marconi16:51
kgriffsflaper87: kk16:52
kgriffs(wrt orient)16:52
kgriffsso, I made this: https://blueprints.launchpad.net/marconi/+spec/alternative-storage-drivers16:52
kgriffsnot something to worry about right now, but maybe later16:52
kgriffsflaper87: so, fwi, Oz is concerned that we can't balance a single queue across multiple storage shards right now16:53
kgriffsbut...16:53
flaper87I really would like us to dedicate some thoughts and time to real queuing technologies and see how we can integrate with those. I haven't dedicated enough time to that myself but I will16:53
kgriffsflaper87: yeah16:53
kgriffsso, the thing is, we keep using these DB things that weren't really designed for queueing, but they work OK16:53
flaper87kgriffs: mmg, why can't we shard queues ?16:53
kgriffsso, there are two things to deal with16:54
kgriffsfirst, the monotonic marker thing16:54
kgriffssecond, behavior of queues16:54
kgriffswrt behavior, you would expect the head and tail of the queue to be the most frequently accessed16:55
kgriffsso, you can't simply do ranged sharding since you just end up with two roaming hotspots16:55
flaper87what about using p_q as sharding key?16:55
kgriffsso, then you say, let's hash it out, so then my writes are fast, but my reads have to hit N nodes to gather in the results16:55
flaper87ah wait, he wants to split the queue16:56
kgriffsyeah16:56
flaper87TBH, I don't think splitting the queue is a good idea.16:56
kgriffsso I believe SQS does something of sharding a single queue but they are probably clever and try to minimize the number of nodes you have to query for a given range16:57
kgriffsflaper87: yeah, that was my next thought16:57
flaper87I don't think there's much benefit from splitting the queue. Keeping it in a single shard makes queries faster, safer and isolated16:57
flaper87helps balancing the load16:57
kgriffsyes, true16:57
*** tedross has quit IRC16:57
kgriffsthe only thing is what to do with a really hot queue16:57
flaper87and more shards can be added without re-balancing others16:57
kgriffsyou can rate-limit to reduce impact on neighbors16:58
kgriffsor you can migrate to less-busy node or eventually to a single-tenant shard, if you can make that cost-effective16:58
flaper87kgriffs: my first thought about that was: "That's the price to pay" which is not useful but kinda true16:58
flaper87a couple of solutions would be to migrate to an isolated node16:58
flaper87add replicas to the shard16:58
kgriffsi think you could have a graduated response16:59
flaper87which will help distributing reads16:59
flaper87right16:59
kgriffsyou have shards that have fewer and fewer queues16:59
kgriffsat some point you get to 1:116:59
kgriffsnow, if the user needs MOR16:59
kgriffsyou must scale up that single shard16:59
kgriffsor just tell them go deploy it yourself16:59
flaper87also, that's one of the reasons why marconi's shards per queue (besides all the complex logics of doing a more granular shard)16:59
kgriffsi'm pretty sure that 99% of users don't need extreme performance out of a single queue17:00
kgriffsflaper87: right, and I'm still not convinced that it is a net win to shard a single queue17:00
kgriffsfor swift it works fine17:00
notmyname?17:01
kgriffsbut queue messages are short-lived and arranged as a time series17:01
flaper87yeah, but swift knows how to do it. Marconi doesn't17:01
kgriffsnotmyname: just talking about sharding object17:01
flaper87I mean, marconi relies on the underlying service17:01
flaper87it doesn't know how to do real sharding17:01
flaper87but anyway, I don't think sharding a queue on mongodb would give any real benefit17:02
flaper87we could do some tests / benchmarks17:02
kgriffsflaper87: yeah, unless you are getting a lot of random access for queue messages, which I don't think is the usual use case!17:02
flaper87exactly17:04
kgriffsi mean, what would you shard on? you could shard on time, but then you have roaming hotspot problem. You could use hash-based sharding but then you end up with scatter-gather queries17:04
kgriffstricky17:04
*** vkmc has quit IRC17:04
flaper87also, splitting the queue has a penalty in the time constraint of marconi's response17:04
kgriffsif you can limit the scatter-gather to a subset of nodes, it could maybe help17:04
kgriffsbut still a P17:04
kgriffsPITA17:04
flaper87we don't have a specific expectation just yet, but we will at some point.17:04
kgriffsflaper87: true17:05
flaper87one more -1 for sharding17:05
kgriffsyou can query multiple nodes in parallel, but then you have to take the time to splice things together17:05
flaper87if makes marconi less fault-tolerant17:05
flaper87if 1 shard goes down, we don't know how much knowledge per queue it has17:05
kgriffsand then you really have to have some kind of proxy that sets the marker which means at some point everything will go through a single choke point anyway17:05
kgriffsflaper87: yep17:06
flaper87it's also true that if a shard goes down, all the info about some queues will be lost17:06
flaper87assuming there are no replicas17:06
kgriffsmakes things very complicated17:06
*** tedross has joined #openstack-marconi17:06
flaper87but anyway, it's about choosing whether to make all clients partially sad or some really sad while others will remain happy17:06
kgriffsgiven all our constraints, i think our best bet is to try the approach of a sort of "winnowing" rebalancer17:06
flaper87all that being said, users can still shard queues, marconi should be able to talk to shard envs easily17:07
kgriffsby users you mean operators?17:07
kgriffsas in, they could have a storage driver that shards itself or relies on underlying db to do it?17:07
* kgriffs is about to run out of battery juice and needs to go borrow an adapter17:09
openstackgerritKurt Griffiths proposed a change to openstack/marconi: fix: Checking whether queue exists adds latency  https://review.openstack.org/6797817:09
flaper87kgriffs: yup17:10
flaper87kgriffs: yup, operator17:10
kgriffskk17:10
kgriffsmakes sense17:10
openstackgerritFlavio Percoco proposed a change to openstack/marconi: Add a _config_options function to Marconi  https://review.openstack.org/6844117:11
kgriffsflaper87: consider, using containers a public provider could give someone a single-tenant marconi instance that rate-limits organically simply by restricting cpu slice they get17:11
openstackgerritKurt Griffiths proposed a change to openstack/marconi: fix(wsgi): Cleanup limit config options  https://review.openstack.org/6759717:12
kgriffswoot17:12
kgriffshttp://tarballs.openstack.org/marconi/17:12
kgriffshttps://github.com/openstack/marconi/tree/milestone-proposed17:13
flaper87w000t17:13
*** zyuan has quit IRC17:14
*** whenry has quit IRC17:15
*** whenry has joined #openstack-marconi17:15
*** tedross has quit IRC17:15
*** alcabrera has joined #openstack-marconi17:16
alcabreraHey! :D17:16
flaper87alcabrera: 'sup man!17:19
flaper87alcabrera: https://github.com/openstack/marconi/tree/milestone-proposed17:20
*** kgriffs is now known as kgriffs_afk17:20
*** chandankumar_ has quit IRC17:24
*** tedross has joined #openstack-marconi17:24
* alcabrera checks17:26
*** vkmc has joined #openstack-marconi17:28
*** chandankumar_ has joined #openstack-marconi17:29
*** jmckind has joined #openstack-marconi17:30
alcabrerawoooo, milestone *cut*17:35
alcabrera:D17:35
alcabreraflaper87: !17:35
flaper87:D17:35
alcabreraso....17:36
alcabrerahmmm17:36
alcabrerawhen do we get to do 'pip install marconi'?17:36
alcabrera:D17:36
alcabrerais that a thing?17:36
alcabrerafor OS projects?17:36
alcabrerabecause marconi would benefit intensely from taht17:36
alcabrera*that17:37
flaper87mmh nope, server modules don't get to pypi, AFAIK.17:37
flaper87but you can pip install -f tarball17:37
alcabrerafair enough17:40
*** amitgandhi has quit IRC18:02
*** amitgandhi has joined #openstack-marconi18:23
*** amitgandhi has quit IRC18:28
*** amitgandhi has joined #openstack-marconi18:29
*** zyuan has joined #openstack-marconi18:33
*** jamieh has quit IRC18:42
*** oz_akan_ has joined #openstack-marconi18:56
*** zyuan has quit IRC19:50
*** cpallares has quit IRC19:50
*** zyuan has joined #openstack-marconi19:52
*** zyuan has joined #openstack-marconi19:53
*** kgriffs_afk is now known as kgriffs20:09
kgriffslovely20:11
kgriffshttp://logs.openstack.org/78/67978/2/check/gate-marconi-python27/7f2fc97/console.html20:11
kgriffsanybody already looking at that?20:11
kgriffsI guess that's a "no"?20:17
kgriffsalcabrera: ^^^20:17
kgriffsanother one: http://logs.openstack.org/97/67597/3/check/gate-marconi-python26/a0f78bc/console.html20:18
kgriffsmalini: ^^^20:18
kgriffsthat either means the wsgi app failed to start (due to unhandled exception) or perhaps we didn't wait long enough for the wsgi server to come up before hitting it?20:20
* alcabrera catches up20:23
alcabrerakgriffs: that's a surprising error20:25
alcabrerano, I haven't even noticed it today20:25
alcabrera'til just now20:25
alcabreraas far as the wsgi app failing to start up, I'm not seeing that `marconi-server` mode from latest master.20:27
kgriffshmm20:31
kgriffsi haven't been able to repro locally20:31
kgriffs(either of those errors)20:31
kgriffsalcabrera: looks like the functional tests don't use marconi-server cmd20:35
kgriffsuses tests.functional.base.MarconiServer instead20:35
*** zyuan has quit IRC20:44
*** malini is now known as malini_afk20:44
*** zyuan has joined #openstack-marconi20:55
*** zyuan has quit IRC20:59
kgriffsbalajiiyer: ping21:06
*** amitgandhi has quit IRC21:07
balajiiyerkgriffs: ^^21:09
*** zyuan has joined #openstack-marconi21:10
*** vkmc has quit IRC21:11
*** amitgandhi has joined #openstack-marconi21:13
*** zyuan has quit IRC21:15
*** amitgandhi has quit IRC21:15
*** amitgandhi has joined #openstack-marconi21:16
*** zyuan has joined #openstack-marconi21:26
*** zyuan has quit IRC21:30
openstackgerritKurt Griffiths proposed a change to openstack/marconi: fix: Move decorator out of class definition since it is static  https://review.openstack.org/6849021:41
*** zyuan has joined #openstack-marconi21:42
kgriffsalcabrera: ^^^ experiment21:44
kgriffsto see if that helps with one of the gate errors we've been seeing21:45
kgriffsok guys, I gotta hit the road.21:45
kgriffsttfn21:45
*** zyuan has quit IRC21:47
alcabrerakgriffs: o/21:51
kgriffsyo21:51
* kgriffs hasn't left quite yet21:51
alcabrerahehe21:52
alcabreraI was about to head out myself.21:52
*** zyuan has joined #openstack-marconi21:58
*** zyuan has quit IRC22:02
*** mpanetta has quit IRC22:03
*** oz_akan_ has quit IRC22:04
*** alcabrera has quit IRC22:04
*** zyuan has joined #openstack-marconi22:14
*** kgriffs is now known as kgriffs_afk22:15
*** zyuan has quit IRC22:18
*** zyuan has joined #openstack-marconi22:30
*** jergerber has joined #openstack-marconi22:31
*** tedross has quit IRC22:32
*** zyuan has quit IRC22:34
*** zyuan has joined #openstack-marconi22:42
*** zyuan has joined #openstack-marconi22:43
*** thomasem has quit IRC22:48
*** balajiiyer has quit IRC22:51
*** amitgandhi has quit IRC23:15
*** jcru has quit IRC23:20
*** zyuan has quit IRC23:21
*** sballe has quit IRC23:22
*** sballe has joined #openstack-marconi23:22
*** jergerber has quit IRC23:24
*** sballe_ has joined #openstack-marconi23:34
*** sballe has quit IRC23:35
*** jmckind has quit IRC23:36
« Tuesday, 2014-01-21 Index Thursday, 2014-01-23 »

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