| *** thorst has joined #openstack-watcher | 00:00 | |
| openstackgerrit | Merged openstack/watcher: Remove true/false return from action.execute() https://review.openstack.org/284541 | 01:13 |
|---|---|---|
| *** vtech has quit IRC | 01:36 | |
| *** thorst has quit IRC | 02:10 | |
| *** thorst has joined #openstack-watcher | 02:11 | |
| *** thorst has quit IRC | 02:19 | |
| *** thorst has joined #openstack-watcher | 03:17 | |
| *** thorst has quit IRC | 03:24 | |
| *** thorst has joined #openstack-watcher | 04:24 | |
| *** thorst has quit IRC | 04:29 | |
| *** jwcroppe has quit IRC | 05:06 | |
| *** thorst has joined #openstack-watcher | 05:26 | |
| *** thorst has quit IRC | 05:34 | |
| *** apoorv has joined #openstack-watcher | 05:52 | |
| apoorv | watcher is trying to use same ports as nova | 06:05 |
| *** thorst has joined #openstack-watcher | 06:31 | |
| *** thorst has quit IRC | 06:38 | |
| *** apoorv has quit IRC | 06:58 | |
| *** apoorv has joined #openstack-watcher | 06:59 | |
| *** jwcroppe has joined #openstack-watcher | 07:04 | |
| *** vtech has joined #openstack-watcher | 07:26 | |
| *** thorst has joined #openstack-watcher | 07:36 | |
| *** vtech has quit IRC | 07:39 | |
| *** thorst has quit IRC | 07:44 | |
| *** apoorv has quit IRC | 07:46 | |
| *** vtech has joined #openstack-watcher | 07:59 | |
| *** vincentfrancoise has joined #openstack-watcher | 08:22 | |
| jed56 | hello apporv | 08:22 |
| jed56 | What is the port ? | 08:23 |
| *** dtardivel has joined #openstack-watcher | 08:23 | |
| jed56 | you can the port in the watcher.conf file | 08:25 |
| jed56 | you will also need to update the watcher keystone endpoints | 08:25 |
| jed56 | this is the port 9322 that is causing a problem ? | 08:26 |
| *** openstackgerrit has quit IRC | 08:33 | |
| *** openstackgerrit has joined #openstack-watcher | 08:33 | |
| *** apoorv has joined #openstack-watcher | 08:37 | |
| *** thorst has joined #openstack-watcher | 08:42 | |
| apoorv | hi vincent: can you give me simple steps to install watcher on a devstack | 08:44 |
| *** thorst has quit IRC | 08:49 | |
| vincentfrancoise | hi apoorv | 09:04 |
| vincentfrancoise | to set up Watcher in DevStack | 09:05 |
| vincentfrancoise | you can follow this piece of doc: https://forge.b-com.com/www/indeed-ccl/doc/watcher/dev/devstack.html | 09:05 |
| jed56 | 9:22 AM <jed56> hello apporv | 09:05 |
| jed56 | 9:23 AM <jed56> What is the port ? | 09:05 |
| jed56 | 9:23 AM → dtardivel joined (uid116506@gateway/web/irccloud.com/x-qcimndcxzldnuacb) | 09:05 |
| jed56 | 9:25 AM <jed56> you can the port in the watcher.conf file | 09:05 |
| jed56 | 9:25 AM <jed56> you will also need to update the watcher keystone endpoints | 09:05 |
| jed56 | 9:26 AM <jed56> this is the port 9322 that is causing a problem ? | 09:05 |
| apoorv | jed56:the ports defined in watcher/docker-compose.yml are conflicting with nova services | 09:18 |
| apoorv | vincent : it is redirecting to https://b-com.com/ | 09:21 |
| vincentfrancoise | I don't know the reason yet | 09:25 |
| vincentfrancoise | but it the meantime, you can generate the doc offline if you whish | 09:25 |
| vincentfrancoise | from the root folder of the watcher git repo | 09:26 |
| vincentfrancoise | issue this command: "tox -e docs" | 09:26 |
| vincentfrancoise | then you will get the whole doc in doc/build/ | 09:26 |
| vincentfrancoise | ok I gave you the wrong link... | 09:32 |
| vincentfrancoise | https://factory.b-com.com/www/watcher/doc/watcher/dev/devstack.html | 09:33 |
| *** thorst has joined #openstack-watcher | 09:47 | |
| *** thorst has quit IRC | 09:54 | |
| *** thorst has joined #openstack-watcher | 10:52 | |
| *** thorst has quit IRC | 10:59 | |
| *** vincentfrancoise has quit IRC | 11:10 | |
| *** jwcroppe has quit IRC | 11:40 | |
| *** thorst has joined #openstack-watcher | 11:47 | |
| *** jwcroppe_ has joined #openstack-watcher | 11:48 | |
| *** jwcroppe_ has quit IRC | 12:22 | |
| *** apoorv has quit IRC | 12:36 | |
| *** jed56 has quit IRC | 13:03 | |
| *** jed56 has joined #openstack-watcher | 13:17 | |
| *** vincentfrancoise has joined #openstack-watcher | 13:25 | |
| *** pkoniszewski has joined #openstack-watcher | 13:25 | |
| sballe_ | Morning | 13:29 |
| vincentfrancoise | Hi | 13:29 |
| pkoniszewski | hello | 13:30 |
| tkaczynski | hi | 13:30 |
| acabot | hi | 13:30 |
| jed56 | hello | 13:31 |
| sballe_ | ok so let's start the discussion around nova live migration and its integration with watcher | 13:32 |
| sballe_ | acabot: do we need to start a meeting to have it logged? | 13:32 |
| acabot | no | 13:32 |
| dtardivel | hi | 13:33 |
| acabot | everything is logged on this channel | 13:33 |
| sballe_ | pkoniszewski: please go ahead | 13:33 |
| pkoniszewski | okay | 13:34 |
| *** jwcroppe has joined #openstack-watcher | 13:34 | |
| pkoniszewski | so during mitaka cycle we implemented a new rest api in nova to expose live migration progress | 13:34 |
| pkoniszewski | i'm thinking about adding another way to expose these details - through, e.g., notificiation | 13:35 |
| pkoniszewski | that might be something that watcher can consume and act in a better way if live migration is stuck or progress is really slow | 13:35 |
| pkoniszewski | question is if it will be useful for watcher at all | 13:36 |
| jed56 | IMHO, it is a interesting for our "applier module" | 13:37 |
| pkoniszewski | i haven't talked with nova folks yet, i need some input to start talking with them, probably on the summit | 13:37 |
| pkoniszewski | but, if u think that rest api is good enough, then we don't need to work on it at all | 13:38 |
| sballe_ | jed56: this would assume we get the actice mode working | 13:38 |
| acabot | when we apply a new optimization strategy on a cluster, we need to get details about whats going on. So if it allows us to follow actions in more detail, it could be great for us | 13:38 |
| pkoniszewski | yeah, this would require active monitoring | 13:39 |
| acabot | I dont think it requires the active mode | 13:39 |
| jed56 | we could easily subscribe to a topic on listen the events | 13:39 |
| acabot | its something we could use in the applier | 13:39 |
| acabot | to check that the migration is still in progress (or not) and then update the action plan | 13:40 |
| acabot | pkoniszewski: did you already started a spec (or kind of) to define the type of event we could get ? | 13:41 |
| tkaczynski | but I guess the question is: do we need "real-time" updates or the pull model (rest api) is good enough | 13:41 |
| pkoniszewski | acabot: no, not yet, i'm trying to organize all the work in LM in Newton before I actually start to write specs | 13:42 |
| acabot | its always better to rely on notifications than pulling the API I guess ;-) | 13:43 |
| sballe_ | +1 | 13:43 |
| pkoniszewski | yeah, i think so, in this case data is very dynami | 13:43 |
| pkoniszewski | acabot: what do you mean by 'the type of event'? | 13:43 |
| jed56 | IMHO, it is always better to receive events compared to pull an API | 13:44 |
| acabot | for now, I would say that the applier implementation is to basic to say that we need this feature, but as the implementation will evolve (I hope in the Newton cycle), we will definitely need it | 13:44 |
| acabot | pkoniszewski: I mean, would it be just "20% of the live migration" is done or do you plan more advanced events ? | 13:45 |
| tkaczynski | "never say never". I'm not saying that pull model is great, just saying that there are different scenarios and usually there are pros and cons in what we are doing | 13:45 |
| sballe_ | acabot: so we'll add a piece that will monitor the events in the applier which will then it turn keeps track of the lifecycle of an action. | 13:46 |
| acabot | tkaczynski: what are the cons about the notifications ? | 13:46 |
| pkoniszewski | acabot: ah okay, so it's very hard to say that 20% of the live migraiton is done, due to dirty pages in each cycle | 13:46 |
| pkoniszewski | acabot: right now the data that we can expose is memory total, transferred and remaining, same for disks in case of block migration | 13:47 |
| jed56 | tkaczynski: I agree :) | 13:47 |
| pkoniszewski | block live migration i mean | 13:47 |
| tkaczynski | acabot: for example when you don't have the infrastructure for consuming the notifications. this needs to be built, stored, maintained. and if the scenario doesn't require real-time data this might be an overkill | 13:48 |
| tkaczynski | I'm just being pragmatic, not religious :) | 13:48 |
| acabot | tkaczynski : but if you use existing messaging system in OpenStack ? | 13:49 |
| acabot | pkoniszewski: you plan to use oslo messaging right ? | 13:49 |
| tkaczynski | ... and since everybody likes the idea about notification (I like it too!), I'll put myself into the devil's advocate shoes, so at least we will consider all the cases | 13:50 |
| pkoniszewski | notifications are already implemented in nova and yes, it uses oslo messaging in background | 13:50 |
| acabot | I would say that it's not a big priority for us but it would be a great improvement for our applier in the future | 13:51 |
| pkoniszewski | good | 13:53 |
| pkoniszewski | do you have some kind of a roadmap defined for the next two cycles? | 13:54 |
| acabot | pkoniszewski: of course if you need reviews on specs and open discussion at the summit, do not hesitate to ping me | 13:54 |
| pkoniszewski | thanks, if i propose this for Newton I will let you know | 13:55 |
| jed56 | pkoniszewski: could please tell me, why did you need this feature ? | 13:55 |
| acabot | pkoniszewski: we are currently building the roadmap for the Newton cycle, I will let you know about it before the summit | 13:57 |
| jed56 | did you need to change the qemu configuration to achieve that ? | 13:57 |
| pkoniszewski | jed56: basically because starting live migration is not enough in many cases, especially if you want to move busy instance, in such case if you don't monitor the process you will end up with never-ending live migration | 13:58 |
| pkoniszewski | acabot: cool, thanks! | 13:59 |
| pkoniszewski | jed56: and we don't need to change qemu | 13:59 |
| pkoniszewski | jed56: LM process can be reconfigured on-the-fly | 13:59 |
| sballe_ | acabot: Are you still here? | 13:59 |
| pkoniszewski | like increase maximum tolerable downtime, increase cpu throtling etc. | 14:00 |
| sballe_ | acabot: how do you want to move forward with he presentation changes? should I take a second stab? | 14:00 |
| pkoniszewski | and the next thing is that soon there will be LM compression enabled in libvirt, but this will require constant monitoring as compression is a cpu-intensive workload | 14:00 |
| jed56 | acabot just left to an another meeting | 14:01 |
| pkoniszewski | so i'm thinking that watcher could take care of, e.g., how many threads are used for compression or compression ratio basing on platform utilization | 14:01 |
| sballe_ | jed56: ok thx | 14:02 |
| jed56 | koniszewski: yes definitely | 14:02 |
| pkoniszewski | i need to drop too as i have nother meeting offline | 14:02 |
| pkoniszewski | in case you have questions you can always ping me on IRC | 14:02 |
| pkoniszewski | thanks for the discussion | 14:02 |
| jed56 | thanks a lot have a nice day | 14:03 |
| jed56 | tkaczynski: i will take a look to your comments today or tomorrow morning | 14:04 |
| jed56 | at | 14:04 |
| tkaczynski | jed56: thanks. I'm updating the spec anyway and will publish it soon. there will be some drawings and boxes :) | 14:04 |
| jed56 | awesome ! :) | 14:05 |
| openstackgerrit | Merged openstack/watcher-specs: New options add into purge command tool https://review.openstack.org/290727 | 14:07 |
| *** jwcroppe has quit IRC | 14:08 | |
| *** openstackgerrit has quit IRC | 14:33 | |
| *** openstackgerrit has joined #openstack-watcher | 14:33 | |
| *** vtech has quit IRC | 15:01 | |
| *** vtech has joined #openstack-watcher | 15:02 | |
| *** jwcroppe has joined #openstack-watcher | 15:05 | |
| acabot | sballe_: i'm back | 15:06 |
| sballe_ | now I am on a call ;-) | 15:06 |
| acabot | sballe_: did you manage to update the slides ? | 15:07 |
| sballe_ | yes I ca edit them but haven't updated yet | 15:07 |
| acabot | ok so you can take a second stab | 15:08 |
| sballe_ | ok will do | 15:08 |
| acabot | I will add a couple of things in the newton roadmap but I'm clear with the "mission" slides | 15:09 |
| acabot | sballe_: could you please review https://review.openstack.org/#/c/291214/ as jwcroppe already did it and I would have it implemented for mitaka | 15:10 |
| sballe_ | will do right after this meeting | 15:11 |
| acabot | thx | 15:12 |
| openstackgerrit | Merged openstack/watcher: Updated action-plugin doc to refer to Voluptuous https://review.openstack.org/295195 | 15:17 |
| *** jwcroppe has quit IRC | 15:39 | |
| acabot | sballe_: and if you can also +1 this fix to add doc8 to watcher-specs https://review.openstack.org/#/c/294051/ ;-) | 15:54 |
| *** jwcroppe has joined #openstack-watcher | 16:36 | |
| *** wootehfoot has quit IRC | 16:37 | |
| *** wootehfoot has joined #openstack-watcher | 16:52 | |
| *** vtech has quit IRC | 17:08 | |
| *** jwcroppe has quit IRC | 17:09 | |
| *** vincentfrancoise has quit IRC | 17:30 | |
| *** vtech has joined #openstack-watcher | 17:42 | |
| *** jwcroppe has joined #openstack-watcher | 17:48 | |
| *** hvprash has joined #openstack-watcher | 18:18 | |
| *** hvprash has quit IRC | 18:32 | |
| *** jed56 has quit IRC | 18:33 | |
| *** dtardivel has quit IRC | 19:48 | |
| *** jwcroppe has quit IRC | 19:54 | |
| *** jwcroppe has joined #openstack-watcher | 19:58 | |
| *** wootehfoot has quit IRC | 20:16 | |
| *** wootehfoot has joined #openstack-watcher | 20:29 | |
| *** jwcroppe has quit IRC | 20:49 | |
| *** jwcroppe has joined #openstack-watcher | 20:53 | |
| *** jwcroppe has quit IRC | 20:58 | |
| *** thorst has quit IRC | 21:00 | |
| *** thorst has joined #openstack-watcher | 21:00 | |
| *** thorst has quit IRC | 21:05 | |
| *** thorst has joined #openstack-watcher | 21:22 | |
| *** thorst has quit IRC | 21:26 | |
| *** wootehfoot has quit IRC | 21:46 | |
| *** openstackgerrit has quit IRC | 21:48 | |
| *** openstackgerrit has joined #openstack-watcher | 21:48 | |
| *** jwcroppe has joined #openstack-watcher | 21:54 | |
| *** wootehfoot has joined #openstack-watcher | 21:59 | |
| *** jwcroppe has quit IRC | 22:02 | |
| *** vtech has quit IRC | 22:14 | |
| *** vtech has joined #openstack-watcher | 22:16 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!