Wednesday, 2015-06-24

« Tuesday, 2015-06-23 Index Thursday, 2015-06-25 »
*** cleverdevil has quit IRC00:06
adam_gdavidlenwell, cool, ill take a closer look at it tomorrow. gotta EOD00:33
davidlenwellyeah .. me too .. my brain is pretty fried00:34
davidlenwelladam_g: catch you tomorrow00:34
openstackgerritSean Roberts proposed stackforge/akanda: update root readme  https://review.openstack.org/19489500:50
openstackgerritSean Roberts proposed stackforge/akanda: Updates outdated repo reference  https://review.openstack.org/19490001:01
*** puranamr has quit IRC01:31
*** sarob has quit IRC01:39
*** puranamr has joined #akanda01:51
*** puranamr has quit IRC01:56
*** puranamr has joined #akanda04:10
*** puranamr has quit IRC05:42
*** puranamr has joined #akanda05:43
*** puranamr has quit IRC05:47
*** puranamr has joined #akanda05:47
*** puranamr has quit IRC06:08
*** openstack has quit IRC08:25
*** openstack has joined #akanda08:26
-cameron.freenode.net- [freenode-info] channel trolls and no channel staff around to help? please check with freenode support: http://freenode.net/faq.shtml#gettinghelp08:26
*** rods has joined #akanda14:12
*** puranamr has joined #akanda14:39
*** puranamr has quit IRC14:43
*** puranamr has joined #akanda14:43
*** puranamr has quit IRC15:06
*** puranamr has joined #akanda15:06
*** puranamr_ has joined #akanda15:32
*** puranam__ has joined #akanda15:36
*** puranamr has quit IRC15:36
*** puranamr_ has quit IRC15:37
*** sarob has joined #akanda15:48
*** puranam__ has quit IRC16:10
*** puranamr has joined #akanda16:10
openstackgerritMerged stackforge/akanda: Updates outdated repo reference  https://review.openstack.org/19490016:11
openstackgerritMerged stackforge/akanda: update root readme  https://review.openstack.org/19489516:11
*** cleverdevil has joined #akanda16:13
adam_gmarkmcclain, you'll want this in any devstack image you snapshot16:41
adam_ghttps://review.openstack.org/#/c/194444/16:41
openstackgerritMerged stackforge/akanda-appliance-builder: remove stale appliance-builder content  https://review.openstack.org/19482516:43
openstackgerritMerged stackforge/akanda-appliance: Add a build script + build_image tox env  https://review.openstack.org/19291516:43
markmcclainadam_g: right... +2/A16:45
openstackgerritMerged stackforge/akanda-rug: Expose RUG API port as a config option  https://review.openstack.org/19444417:11
ryanpetrelloI had an akanda event management thought17:19
ryanpetrelloright now all events in the queue are basically given fifo priority17:19
ryanpetrellobut I wonder if aliveness/health checks should have a lower priority17:19
ryanpetrelloe.g., "there's a new router that needs to be booted" or "a floating IP was just created" are probably higher priority then "are all x1000 routers responsive?"17:19
ryanpetrellowe get into a situation now where the health subprocess blasts the queue with thousands of health events every few minutes17:20
ryanpetrelloand it takes the rug a bit to catch up17:20
ryanpetrelloin the meantime, other user-driven actions have to wait17:21
adam_gryanpetrello, that sounds reasonable. the health check is part of the POLL event, right?17:22
ryanpetrelloyep17:23
adam_gmarkmcclain, not sure what inconsistency you're refering to in first comment @ https://review.openstack.org/#/c/194841/2/akanda/rug/test/unit/api/test_nova_wrapper.py17:23
ryanpetrelloit would probably not be a big deal to make the event queue be some type of priority queue17:23
ryanpetrelloand just have the POLL event be lower priority17:23
markmcclainadam_g: oops that should have been the line below17:23
markmcclains/Equals/Equal/17:23
adam_gah right17:24
openstackgerritAdam Gandelman proposed stackforge/akanda-rug: Fix ability to recover from an existing appliance VM  https://review.openstack.org/19484117:24
markmcclainI know it's totally unrelated by noticed it when I was checking to see how we were comparing none's in the rest of the file17:24
adam_gyea17:25
adam_gi think theres a linter check that will catch that one17:25
*** puranamr has quit IRC17:26
markmcclainI think there is too17:26
*** puranamr has joined #akanda17:26
*** puranamr has quit IRC17:30
*** puranamr has joined #akanda17:30
adam_ghmm patch bomb heads up, i need to backport a buncha akanda-rug devstack changes to stable/kilo + juno17:31
adam_g...to get it functional in the gate.17:31
*** openstackgerrit has quit IRC17:38
*** openstackgerrit has joined #akanda17:38
openstackgerritMerged stackforge/akanda-rug: Capture and gracefully handle failed interface attach/detach API calls.  https://review.openstack.org/19329217:53
adam_gugh, actually this is going to be a PITA to get working. might make more sense to disable devstack runs against kilo+juno17:54
adam_gdavidlenwell, did a first pass on that driver patch18:18
davidlenwellthanks.. im gona post the patches that split it up in a bit18:18
adam_galso, is 'instance' the best thing to replace 'vm' with?18:18
adam_g...if we're talking about supporting containers in the future18:19
davidlenwellits what mark and I came up with while chatting about it18:19
davidlenwellseems to cover containers, vms, and bare metal18:19
adam_gif we end up doing application containers in the future, those would be hosted on instances, so there'd be some confusion there again18:20
davidlenwell;hmm .. well I am open to suggestions18:20
adam_gi dont have any better ideas, just seems to be overlapping with nova terms18:21
adam_gmarkmcclain, ?18:21
adam_gor we could just punt and cross that bridge later18:21
adam_gbrb, walking home18:21
davidlenwellyeah .. I'm thinking that18:21
markmcclainyeah around18:21
davidlenwellmarkmcclain: we are just talking about vm vs instance18:21
markmcclainah18:22
davidlenwelladam_g is concerned about confusion18:23
markmcclainyeah... it's the openstack way to confuse terms18:23
davidlenwellshould I just punt on naming?  it came up because I wanted to call "router_id" something else18:23
markmcclainwe could have called it a project/tenant/group18:23
markmcclainor policy :)18:23
markmcclainat one time I was thinking we could use resource_id for logical resources18:24
davidlenwellcould just call it "thing"18:24
markmcclainthe only issue is that doesn't tell you want that thing represents18:24
davidlenwellmarkmcclain: thats what I had before we talked the other day18:24
markmcclainfor IDs for a logical resource then resource_id makes sense18:25
markmcclainfor the other side where we're managing an instance of a logical thing then that's where seems like instance_(id|info) could be universal18:27
markmcclainand let the driver interpret18:27
davidlenwellsee I still like instance18:28
davidlenwellonce I moved everything over to that it all clicked18:28
sarobi think we are using instance in the correct way18:29
sarobthe nova overlapping term, is a good thing18:29
cleverdevilwhat about "service"18:29
cleverdevilor "provider"18:29
davidlenwelldoh18:29
cleverdevilor "backend"18:29
davidlenwellhow about junior18:30
davidlenwell;)18:30
cleverdevilheh18:30
sarobugh18:30
sarobservice backend sounds dirty18:30
cleverdevilindeed :)18:30
sarobservice provider sounds less clear than service instance18:31
cleverdevilyeah, probably18:31
markmcclainservice is already really overloaded in OpenStack18:33
adam_gsarob, the overlapping is not a good thing if, in the future, an instance is actually a container running on an instance, which is hosting other containers. thats my main concern18:33
adam_gbut actually i think we can punt on this till the nodepooling/pez thing progresses and tackle it there?18:33
sarobi think matching nova terms right now will make it easier for the short term18:34
cleverdevilhuh... what about node?18:34
cleverdevil:)18:34
cleverdevilits more generic18:34
sarobsense davidlenwell has started it already18:34
cleverdeviland not overloaded18:34
cleverdevilthough, instance is fine for now, IMO18:35
cleverdevilI do think I like node a little better...18:35
sarobcleverdevil: im not against it18:35
sarobbut id like to follow nova18:35
sarobrather come up with our own18:35
adam_gcleverdevil, node is an ironic thing :)18:35
sarobadam_g: ugh!18:35
cleverdevilsure18:35
cleverdevilyeah, but that's fine18:35
cleverdevilnode = bare metal node18:35
sarobyeah18:36
cleverdevilnode could just as easily be a container node18:36
cleverdevilor a vm node18:36
sarobmy head hurt18:36
sarobs18:36
cleverdevilbut, the same could potentially be said for instance18:36
cleverdeviland right now, all we do is VMs18:36
cleverdevilso, instance makes perfect sense today18:36
sarob+118:36
markmcclainyeah we can stick with instances for now until a better term pops up18:37
davidlenwellI like node18:37
davidlenwellbut instance is better and the works already mostly done18:37
adam_gok cool18:37
* cleverdevil nods18:37
cleverdevilwfm!18:37
sarobinstance sounds like the all around best choice18:38
adam_g... for now18:38
davidlenwell;) adam_g18:38
saroblunch tine!!!!18:39
*** puranamr has quit IRC18:41
*** puranamr has joined #akanda18:42
* markmcclain wanders off to find lunch18:45
*** puranamr has quit IRC18:50
*** puranamr has joined #akanda18:50
openstackgerritAdam Gandelman proposed stackforge/akanda-rug: Fix ability to recover from an existing appliance VM  https://review.openstack.org/19484118:53
ryanpetrelloguys guys, what about "computational device"19:10
*** cleverde_ has joined #akanda20:02
*** puranamr_ has joined #akanda20:04
*** puranamr has quit IRC20:04
*** cleverde_ has quit IRC20:04
*** cleverde_ has joined #akanda20:05
*** cleverdevil has quit IRC20:05
*** cleverde_ has quit IRC20:11
*** cleverdevil has joined #akanda20:12
davidlenwellryanpetrello: lol20:13
openstackgerritDoug Hellmann proposed stackforge/akanda-rug: Drop use of 'oslo' namespace package  https://review.openstack.org/19530420:49
*** puranamr_ has quit IRC21:12
*** cleverde_ has joined #akanda21:16
*** puranamr has joined #akanda21:18
*** cleverdevil has quit IRC21:19
davidlenwelldhellmann: nice patch :)21:40
*** puranamr has quit IRC22:00
*** puranamr has joined #akanda22:02
openstackgerritMerged stackforge/akanda-rug: Drop use of 'oslo' namespace package  https://review.openstack.org/19530422:05
*** puranamr has quit IRC22:05
*** puranamr has joined #akanda22:13
*** puranamr has quit IRC22:15
*** puranamr has joined #akanda22:18
*** puranamr has quit IRC22:19
*** puranamr has joined #akanda22:19
adam_gryanpetrello, so i might have another use case for event prio-- we talked about using a distributed hash table to distribute routers among a cluster of RUGs22:45
adam_gim playing with that now, adding a new rebalance event that happens when members come/go, which would need priority22:46
*** puranamr has quit IRC22:53
*** rods has quit IRC22:56
*** puranamr_ has joined #akanda23:00
*** puranamr_ has quit IRC23:04
*** puranamr has joined #akanda23:22
*** puranamr_ has joined #akanda23:34
*** puranamr has quit IRC23:38
*** puranamr_ has quit IRC23:40
*** puranamr has joined #akanda23:53
« Tuesday, 2015-06-23 Index Thursday, 2015-06-25 »

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