| *** tjakobs has joined #openstack-powervm | 00:01 | |
| *** tjakobs has quit IRC | 00:06 | |
| *** thorst has quit IRC | 00:06 | |
| *** thorst has joined #openstack-powervm | 01:04 | |
| *** thorst has quit IRC | 01:11 | |
| *** thorst has joined #openstack-powervm | 01:15 | |
| *** thorst has quit IRC | 01:16 | |
| *** esberglu has quit IRC | 01:43 | |
| *** esberglu has joined #openstack-powervm | 02:14 | |
| *** thorst has joined #openstack-powervm | 02:17 | |
| openstackgerrit | Eric Berglund proposed openstack/nova-powervm: DNM: ci check https://review.openstack.org/328315 | 02:22 |
|---|---|---|
| openstackgerrit | Eric Berglund proposed openstack/nova-powervm: DNM: CI Check2 https://review.openstack.org/328317 | 02:22 |
| openstackgerrit | Eric Berglund proposed openstack/nova-powervm: DNM1 https://review.openstack.org/300231 | 02:22 |
| openstackgerrit | Eric Berglund proposed openstack/nova-powervm: DNM: CI Check https://review.openstack.org/295935 | 02:22 |
| openstackgerrit | Eric Berglund proposed openstack/nova-powervm: DNM: Test Change Set 2 https://review.openstack.org/300232 | 02:22 |
| *** kriskend has quit IRC | 02:24 | |
| *** kriskend_ has quit IRC | 02:24 | |
| *** thorst has quit IRC | 02:25 | |
| *** esberglu has quit IRC | 02:35 | |
| *** thorst has joined #openstack-powervm | 02:43 | |
| *** thorst has quit IRC | 02:43 | |
| *** tjakobs has joined #openstack-powervm | 02:53 | |
| *** esberglu has joined #openstack-powervm | 02:58 | |
| *** esberglu has quit IRC | 02:59 | |
| *** tjakobs has quit IRC | 03:00 | |
| *** Ashana has joined #openstack-powervm | 03:35 | |
| *** Ashana has quit IRC | 03:39 | |
| *** Ashana has joined #openstack-powervm | 03:41 | |
| *** thorst has joined #openstack-powervm | 03:44 | |
| *** Ashana has quit IRC | 03:45 | |
| *** Ashana has joined #openstack-powervm | 03:46 | |
| *** Ashana has quit IRC | 03:52 | |
| *** thorst has quit IRC | 03:52 | |
| *** esberglu has joined #openstack-powervm | 03:55 | |
| *** kotra03 has joined #openstack-powervm | 03:56 | |
| *** esberglu has quit IRC | 04:02 | |
| *** thorst has joined #openstack-powervm | 04:51 | |
| *** thorst has quit IRC | 04:58 | |
| *** jwcroppe has joined #openstack-powervm | 05:11 | |
| *** jwcroppe has quit IRC | 05:51 | |
| *** thorst has joined #openstack-powervm | 05:56 | |
| *** thorst has quit IRC | 06:03 | |
| *** thorst has joined #openstack-powervm | 07:01 | |
| *** thorst has quit IRC | 07:08 | |
| *** Ashana has joined #openstack-powervm | 07:14 | |
| *** Ashana has quit IRC | 07:18 | |
| *** openstackgerrit has quit IRC | 08:03 | |
| *** openstackgerrit has joined #openstack-powervm | 08:03 | |
| *** thorst has joined #openstack-powervm | 08:08 | |
| *** alechevalier has joined #openstack-powervm | 08:09 | |
| *** alechevalier has quit IRC | 08:11 | |
| *** k0da has joined #openstack-powervm | 08:13 | |
| *** thorst has quit IRC | 08:13 | |
| *** thorst has joined #openstack-powervm | 09:10 | |
| *** thorst has quit IRC | 09:18 | |
| *** kotra03 has quit IRC | 09:25 | |
| openstackgerrit | Ravi Kumar Kota proposed openstack/nova-powervm: WIP: PowerVM: Implement get_instance_diagnostics https://review.openstack.org/350525 | 09:40 |
| *** kotra03 has joined #openstack-powervm | 10:02 | |
| *** thorst has joined #openstack-powervm | 10:16 | |
| *** thorst has quit IRC | 10:23 | |
| *** kotra03 has quit IRC | 10:25 | |
| *** kotra03 has joined #openstack-powervm | 10:25 | |
| *** thorst has joined #openstack-powervm | 11:20 | |
| *** svenkat has joined #openstack-powervm | 11:26 | |
| *** thorst has quit IRC | 11:28 | |
| *** smatzek has joined #openstack-powervm | 11:29 | |
| openstackgerrit | Arun Mani proposed openstack/nova-powervm: Deploy of a VM occasionally fails due to invalid keystone token https://review.openstack.org/351270 | 11:31 |
| openstackgerrit | Arun Mani proposed openstack/nova-powervm: Deploy of a VM occasionally fails due to invalid keystone token https://review.openstack.org/351270 | 11:32 |
| *** thorst has joined #openstack-powervm | 11:48 | |
| *** Ashana has joined #openstack-powervm | 12:04 | |
| *** Ashana has quit IRC | 12:28 | |
| *** thorst_ has joined #openstack-powervm | 12:45 | |
| *** thorst has quit IRC | 12:49 | |
| *** catintheroof has quit IRC | 12:53 | |
| *** catintheroof has joined #openstack-powervm | 13:05 | |
| openstackgerrit | ymadhavi@in.ibm.com proposed openstack/nova-powervm: [WIP] Plug updated VIFs https://review.openstack.org/350768 | 13:10 |
| *** esberglu has joined #openstack-powervm | 13:12 | |
| *** madhaviy has joined #openstack-powervm | 13:12 | |
| *** mdrabe has joined #openstack-powervm | 13:14 | |
| *** lmtaylor has joined #openstack-powervm | 13:18 | |
| thorst_ | madhaviy: Great update on the WIP. I think the next step is to update the tests to validate the new code that was put in | 13:32 |
| thorst_ | the iteration you did now got it to pass the existing tests, which is important. Next though is validation of the new bits. | 13:32 |
| *** jwcroppe has joined #openstack-powervm | 13:33 | |
| madhaviy | ok, I will add more tests | 13:34 |
| *** kriskend has joined #openstack-powervm | 13:38 | |
| *** kriskend_ has joined #openstack-powervm | 13:38 | |
| *** seroyer has joined #openstack-powervm | 13:40 | |
| *** tjakobs has joined #openstack-powervm | 13:46 | |
| *** seroyer has quit IRC | 13:51 | |
| *** catintheroof has quit IRC | 14:00 | |
| *** burgerk has joined #openstack-powervm | 14:01 | |
| *** openstackgerrit has quit IRC | 14:03 | |
| *** openstackgerrit has joined #openstack-powervm | 14:03 | |
| *** edmondsw has joined #openstack-powervm | 14:04 | |
| *** catintheroof has joined #openstack-powervm | 14:07 | |
| svenkat | @efried: I have a suggestion for redundancy and capacity data in upcoming SR-IOV vif driver | 14:36 |
| efried | tell me | 14:36 |
| svenkat | if you can make them two separate methods and let them return the required data | 14:36 |
| efried | Not sure what you're referencing. | 14:37 |
| thorst_ | efried: at some point, would like to chat about how to detect when the port labels change from networking-powervm | 14:44 |
| thorst_ | I'm wondering if we can use an event model instead of polling? | 14:44 |
| efried | Well | 14:44 |
| efried | I put up a patch set last night that kinda does both. | 14:45 |
| efried | The rpc_loop re-grabs the data | 14:45 |
| efried | But the existing (inherited) event handler posts the changes back to the world. | 14:45 |
| efried | I'm still trying to get my brain around the order of operations. | 14:46 |
| efried | thorst_, svenkat: does this vNIC plug/unplug stuff need to be able to work for both LPAR and VIOS types? | 14:49 |
| thorst_ | efried: LPARs | 14:49 |
| thorst_ | efried: I'll look at the patch set. I'm worried about getting that data every 2 seconds. | 14:49 |
| efried | k, makes life a little easier. | 14:49 |
| thorst_ | efried: well, to be clear, plug/unplug only works against LPARs because OpenStack manages client VMs | 14:50 |
| thorst_ | we should never have a client VM that is a VIOS type | 14:50 |
| efried | That's kinda what I figured. | 14:50 |
| thorst_ | unless apearson goes crazy | 14:50 |
| efried | As far as getting the data every 2s - it's a very quick call, and it's happening per vif plugin (and only for the SR-IOV one at that) - not for like every instance or anything. | 14:51 |
| thorst_ | ahhh | 14:51 |
| thorst_ | I should review code before I complain about it | 14:51 |
| efried | I take your point, and could certainly be persuaded that it should be event-driven. It's just two orders of magnitude more complex that way, so there would have to be a compelling reason. | 14:52 |
| thorst_ | well, I do personally hate event only models | 14:53 |
| thorst_ | cause an event always slips through the cracks | 14:53 |
| *** lmtaylor has quit IRC | 14:53 | |
| *** mdrabe has quit IRC | 14:55 | |
| efried | thorst_, agree. Always seem really brittle. | 14:57 |
| efried | Trying to think if etag would save us anything. | 14:58 |
| efried | I can cache the System wrapper and use the refresh() method rather than getting a new one each time. That's better code - but I'm not sure it's any more efficient. | 14:58 |
| *** mdrabe has joined #openstack-powervm | 15:02 | |
| *** dwayne has joined #openstack-powervm | 15:05 | |
| *** lmtaylor has joined #openstack-powervm | 15:07 | |
| thorst_ | efried: not more efficient...but yeah | 15:07 |
| efried | It may be a trivial amount more efficient because the REST response has no XML in it. | 15:07 |
| efried | ...if nothing has changed. | 15:07 |
| openstackgerrit | Eric Fried proposed openstack/nova-powervm: WIP: VIF driver implementation for SR-IOV https://review.openstack.org/343419 | 15:08 |
| efried | thorst_, svenkat ^^ like this, for redundancy/capacity? | 15:08 |
| openstackgerrit | Eric Fried proposed openstack/networking-powervm: WIP: Mechanism driver for powervm SR-IOV interface https://review.openstack.org/343423 | 15:09 |
| svenkat | efried: neutron port id shold be involved here. there could be more than one SR-IOV adapater attached to vif. extra specs is for entire instance | 15:11 |
| efried | So the consumer would have to make the extra_spec entry a dict | 15:12 |
| efried | I assume that's... possible? | 15:12 |
| svenkat | yes… i am getting that dictionary details… one moment | 15:12 |
| efried | '<physnet>': { 'redundancy': 3, 'capacity': 0.05 } | 15:12 |
| efried | ? | 15:12 |
| *** jwcroppe has quit IRC | 15:12 | |
| *** jwcroppe has joined #openstack-powervm | 15:13 | |
| efried | Or even { '<pport_loc>': { 'capacity': 0.05 }, ..., '<physnet_name>' { 'redundancy': 3, 'capacity': 0.07 }, ... } | 15:14 |
| svenkat | {'powervc-sriov-ports':[{'port_uuid': 'ddc24826-504d-422b-b758-a84bcbe77992', 'capacity' : '50', 'redundancy-level': '4'}]} | 15:15 |
| efried | Where did that come from? | 15:15 |
| efried | Where does this get created? By the admin when defining the flavor? How did they know the port UUID? | 15:16 |
| svenkat | you need to create the port before you deploy for ‘direct’ types , so you know the port id | 15:16 |
| efried | I don't think so. | 15:16 |
| svenkat | otherwise, ‘normal’ port is getting created if you specify network id during boot | 15:17 |
| efried | thorst_, help me out here. There should be no need to do this in two steps, right? You just have to select the physnet name. | 15:17 |
| efried | Say wha? Wouldn't that be a configuration thing? | 15:18 |
| thorst_ | efried: I think that 1) that is a powervm thing (not powervc). 2) you probably want to set capacity and redundancy per VIF. Not per physical network | 15:19 |
| thorst_ | different workloads on the same physical network have different needs... | 15:19 |
| efried | thorst_, would we generally create only one vif per physnet per instance? | 15:27 |
| thorst_ | efried: I think so. Generally | 15:32 |
| *** burgerk has quit IRC | 15:36 | |
| *** lmtaylor has left #openstack-powervm | 15:50 | |
| thorst_ | esberglu: it seems that quota_floating_ips can no longer be set to -1. I think this may actually be a bug you want to open to nova | 16:12 |
| thorst_ | I'm seeing that in new devstacks and I know our undercloud uses it | 16:12 |
| esberglu | thorst: Yeah I hit that too. Do we even need that config option? There are a bunch of other config options that had min=0 added, while others had min=-1 added in this change. Not sure if they were intentionally doing it that way | 16:18 |
| esberglu | https://github.com/openstack/nova/commit/2eebb5edc6c236252e2da1ab1a3e0120de160acd | 16:18 |
| esberglu | I don’t know why they would get rid disabling some of the quotas but since its not all of them it seems to be on purpose | 16:25 |
| *** burgerk has joined #openstack-powervm | 16:27 | |
| thorst_ | well, it seems like a proper bug | 16:36 |
| thorst_ | but we can probably toss that one | 16:36 |
| *** k0da has quit IRC | 16:36 | |
| esberglu | thorst_: Okay I will open a bug for it | 16:39 |
| thorst_ | and maybe even a good bug fix opportunity | 16:39 |
| esberglu | Yep | 16:42 |
| *** lmtaylor has joined #openstack-powervm | 16:55 | |
| openstackgerrit | Eric Fried proposed openstack/nova-powervm: WIP: VIF driver implementation for SR-IOV https://review.openstack.org/343419 | 16:55 |
| *** madhaviy has quit IRC | 16:59 | |
| *** kotra03 has quit IRC | 17:06 | |
| thorst_ | efried: https://github.com/openstack/networking-powervm/blob/master/doc/source/devref/usage.rst | 17:52 |
| thorst_ | I think that is the conf option doc | 17:52 |
| thorst_ | we'd need that updated for sr-iov too | 17:52 |
| *** seroyer has joined #openstack-powervm | 18:04 | |
| efried | thorst_, that guy is pretty sparse. E.g. it doesn't tell you *how* to set up your SEA, or even provide a pointer to a doc that does. | 18:32 |
| *** kriskend has quit IRC | 18:35 | |
| *** kriskend_ has quit IRC | 18:35 | |
| efried | For SR-IOV, I was thinking we would need to document | 18:36 |
| efried | - the pvmctl command to assign port labels | 18:36 |
| efried | - the neutron port-create command syntax (including the redundancy & capacity stuff we discussed) | 18:36 |
| efried | - the conf options: seems like we need to document the new ml2_conf.ini mechanism_drivers value ('pvm_sriov') and the new nova.conf setting ('vnic_required_vfs') | 18:36 |
| efried | Anything else? | 18:36 |
| thorst_ | efried: I think things like how to boot with a SR-IOV based vnic | 18:52 |
| thorst_ | but I guess that's in the neutron port-create | 18:52 |
| efried | We gotta assume they know how to specify an existing port to nova boot. | 18:52 |
| *** esberglu has quit IRC | 18:53 | |
| *** catintheroof has quit IRC | 19:00 | |
| efried | thorst_ - "These configuration options go in the AGENT section of the CONF file." | 19:01 |
| efried | Which CONF file?? | 19:01 |
| thorst_ | ml2.conf | 19:09 |
| *** esberglu has joined #openstack-powervm | 19:10 | |
| efried | thorst_ A section literally called [AGENT] ? | 19:11 |
| efried | or a section called [pvm_sriov] ? | 19:12 |
| thorst_ | I think literally called [AGENT] | 19:17 |
| thorst_ | hop on neo38 | 19:17 |
| efried | Well. | 19:19 |
| efried | Is heal_and_optimize_interval a neutron thing? | 19:20 |
| thorst_ | efried: its a me thing | 19:21 |
| efried | Is it a private/undocumented conf option? | 19:21 |
| efried | I mean, it's undocumented; is it supposed to be undocumented? | 19:21 |
| *** thorst__ has joined #openstack-powervm | 19:23 | |
| thorst__ | no, it was meant to be documented. | 19:23 |
| efried | thorst__, I'm going to open a bug, since a) I don't want to fold that into my current change set; b) I wouldn't know how to document it anyway; and c) I don't know if there are more opts that need to be documented. | 19:25 |
| *** burgerk_ has joined #openstack-powervm | 19:25 | |
| efried | thorst__ - I suppose it's too late to rename bridge_mappings to sea_bridge_mappings | 19:25 |
| thorst__ | correct. | 19:26 |
| *** thorst_ has quit IRC | 19:26 | |
| *** thorst__ is now known as thorst_ | 19:26 | |
| *** burgerk has quit IRC | 19:29 | |
| efried | svenkat: I shouldn't need to say --binding:profile type=dict physical_network=whatever | 19:32 |
| efried | The physical network name is already in the vif | 19:32 |
| efried | am I wrong? | 19:32 |
| svenkat | i found that only when you do —binding:profile type-dict physical_network=xxx i could to vif.get_physical_network() and get it… | 19:39 |
| svenkat | otherwise i was getting none. | 19:39 |
| svenkat | or did you change anything in mechnism driver to get this to work? or actually you can also get physical network from segment (neutron) and put it in vif_details (get_vif_details) in mech driver code | 19:40 |
| efried | svenkat, looking at get_physical_network(), it should just work by getting the network name out of ['network']['meta']['physical_network'] | 19:41 |
| efried | I'll try it out. | 19:41 |
| svenkat | ok.. | 19:41 |
| svenkat | so we can use binding:profile only for redundancy and capacity. .. | 19:41 |
| efried | right, that was my thought. | 19:44 |
| *** k0da has joined #openstack-powervm | 19:44 | |
| *** erlarese has quit IRC | 19:45 | |
| efried | thorst_ https://bugs.launchpad.net/networking-powervm/+bug/1610404 | 19:56 |
| openstack | Launchpad bug 1610404 in networking-powervm "Update config options table in usage.rst" [Undecided,New] | 19:56 |
| efried | thanks, openstack bot. I win. | 19:56 |
| efried | thorst_, svenkat: Should we put the redundancy setting into ml2_conf.ini instead? | 19:57 |
| efried | Heck, should we even bother with the conf setting, now that we have a way to specify it in the port-create? | 19:58 |
| svenkat | efreid: yes. it should be in port-create . | 19:58 |
| efried | um | 19:59 |
| efried | It's going to be in port-create for sure. | 19:59 |
| efried | The question is whether we should bother having a conf option for it anymore | 19:59 |
| efried | and if so, whether it should be in the nova.conf (as currently designed) or if it makes more sense for it to be in the ml2_conf.ini | 19:59 |
| efried | I know it's being *used* in the vif driver, but from the user's perspective, it may make more sense as a ml2 plugin conf option. | 20:00 |
| thorst_ | efried: I like your suggestion. | 20:01 |
| efried | which one? | 20:01 |
| thorst_ | nova can always assume it comes from one place that way | 20:01 |
| thorst_ | put it in ml2. | 20:01 |
| efried | okay. | 20:01 |
| efried | Should we put a capacity default in there too? | 20:01 |
| thorst_ | yep | 20:01 |
| thorst_ | totally dig that | 20:02 |
| efried | Second thoughts... not sure we can reasonably do that. | 20:02 |
| thorst_ | (until I don't) | 20:02 |
| efried | Cause it has to be a multiple of the min capacity granularity. | 20:02 |
| thorst_ | well, we could reasonably set it to 2% | 20:02 |
| thorst_ | but... | 20:02 |
| efried | And it may be different for different ports | 20:02 |
| efried | for different groups of ports | 20:02 |
| efried | for different networks. | 20:02 |
| thorst_ | right...if it were ever to go down to 1% | 20:02 |
| efried | too complex to handle. | 20:03 |
| efried | leave it out. | 20:03 |
| thorst_ | well, way to build someone up and then destroy their hopes | 20:06 |
| openstackgerrit | Drew Thorstensen proposed openstack/networking-powervm: Simplify host_uuid and gets https://review.openstack.org/344399 | 20:06 |
| *** thorst__ has joined #openstack-powervm | 20:16 | |
| *** apearson has joined #openstack-powervm | 20:17 | |
| *** thorst_ has quit IRC | 20:18 | |
| *** svenkat has quit IRC | 20:19 | |
| *** apearson has quit IRC | 20:31 | |
| *** thorst__ has quit IRC | 20:46 | |
| *** burgerk_ has quit IRC | 20:49 | |
| *** thorst_ has joined #openstack-powervm | 20:50 | |
| *** smatzek has quit IRC | 20:54 | |
| *** thorst_ has quit IRC | 20:55 | |
| *** lmtaylor has left #openstack-powervm | 20:56 | |
| openstackgerrit | Eric Fried proposed openstack/nova-powervm: WIP: VIF driver implementation for SR-IOV https://review.openstack.org/343419 | 21:02 |
| openstackgerrit | Eric Fried proposed openstack/nova-powervm: WIP: VIF driver implementation for SR-IOV https://review.openstack.org/343419 | 21:12 |
| openstackgerrit | Eric Fried proposed openstack/nova-powervm: WIP: VIF driver implementation for SR-IOV https://review.openstack.org/343419 | 21:18 |
| *** dwayne has quit IRC | 21:30 | |
| *** edmondsw has quit IRC | 21:42 | |
| *** mdrabe has quit IRC | 22:05 | |
| *** tjakobs has quit IRC | 22:21 | |
| openstackgerrit | Eric Fried proposed openstack/networking-powervm: WIP: Mechanism driver for powervm SR-IOV interface https://review.openstack.org/343423 | 22:25 |
| *** seroyer has quit IRC | 22:34 | |
| *** k0da has quit IRC | 23:37 | |
| *** seroyer has joined #openstack-powervm | 23:45 | |
| *** esberglu has quit IRC | 23:52 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!