| devananda | jroll: yes. we do need log shipping from the agent | 00:24 |
|---|---|---|
| devananda | adam_g: i'd be fine with a deploy-debug element or some such | 00:24 |
| devananda | adam_g: worth pinging in tripleo, if you haven't alraedy | 00:24 |
| * devananda looks at the pile of reviews waiting for him | 00:25 | |
| devananda | starting with 77623 | 00:26 |
| *** matsuhashi has joined #openstack-ironic | 00:26 | |
| jroll | devananda: right | 00:26 |
| comstud | russell_h: just the logging module in oslo | 00:29 |
| comstud | or rather, 'log' | 00:30 |
| devananda | comstud: got a sec ? | 00:34 |
| devananda | nvm... bbiab | 00:38 |
| *** matty_dubs|gone has quit IRC | 00:43 | |
| *** matty_dubs|gone has joined #openstack-ironic | 00:50 | |
| *** derekh has quit IRC | 00:50 | |
| *** dividehex has quit IRC | 00:57 | |
| *** zigo has quit IRC | 00:58 | |
| *** jrist has quit IRC | 00:58 | |
| *** GheRivero has quit IRC | 00:58 | |
| *** dividebin has joined #openstack-ironic | 00:58 | |
| *** GheRivero has joined #openstack-ironic | 00:58 | |
| *** jrist has joined #openstack-ironic | 00:58 | |
| *** openstack has joined #openstack-ironic | 01:05 | |
| *** russell_h has joined #openstack-ironic | 01:05 | |
| *** lazy_prince has joined #openstack-ironic | 01:07 | |
| *** lazy_prince is now known as killer_prince | 01:07 | |
| openstackgerrit | A change was merged to openstack/ironic: Add TaskManager lock on change port data https://review.openstack.org/77623 | 01:07 |
| *** dwalleck has joined #openstack-ironic | 01:10 | |
| *** dwalleck_ has joined #openstack-ironic | 01:11 | |
| *** eghobo has quit IRC | 01:13 | |
| *** dwalleck has quit IRC | 01:14 | |
| *** zigo has quit IRC | 01:27 | |
| *** zigo has joined #openstack-ironic | 01:33 | |
| *** nosnos has joined #openstack-ironic | 01:39 | |
| *** matsuhashi has quit IRC | 01:58 | |
| *** hemna has quit IRC | 02:00 | |
| *** vkozhukalov has joined #openstack-ironic | 02:07 | |
| Shrews | adam_g: awesome. i'll give that a try in the morning | 02:11 |
| *** matsuhashi has joined #openstack-ironic | 02:17 | |
| *** zigo has quit IRC | 02:19 | |
| *** zigo has joined #openstack-ironic | 02:23 | |
| *** killer_prince is now known as lazy_prince | 02:29 | |
| *** vkozhukalov has quit IRC | 02:36 | |
| *** harlowja is now known as harlowja_away | 02:48 | |
| *** rloo has quit IRC | 02:50 | |
| *** rsacharya has joined #openstack-ironic | 02:53 | |
| *** rameshg87 has joined #openstack-ironic | 03:15 | |
| *** rwsu has quit IRC | 03:19 | |
| *** matsuhashi has quit IRC | 03:23 | |
| *** eghobo has joined #openstack-ironic | 03:45 | |
| *** rwsu has joined #openstack-ironic | 03:50 | |
| *** killer_prince has joined #openstack-ironic | 04:10 | |
| *** matsuhashi has joined #openstack-ironic | 04:25 | |
| *** rameshg87 has quit IRC | 04:29 | |
| *** rameshg87 has joined #openstack-ironic | 05:02 | |
| *** rameshg87 has left #openstack-ironic | 05:03 | |
| *** rsacharya has quit IRC | 05:34 | |
| *** loki_ has joined #openstack-ironic | 05:38 | |
| *** vkozhukalov has joined #openstack-ironic | 05:53 | |
| comstud | devananda: sorry, I bailed... I'll be back tomorrow :) | 06:02 |
| openstackgerrit | Jenkins proposed a change to openstack/ironic: Imported Translations from Transifex https://review.openstack.org/78862 | 06:11 |
| *** dwalleck_ has quit IRC | 06:19 | |
| *** romcheg1 has joined #openstack-ironic | 06:40 | |
| *** saju_m has joined #openstack-ironic | 06:52 | |
| *** romcheg1 has quit IRC | 06:52 | |
| *** bashok has joined #openstack-ironic | 06:55 | |
| adam_g | devananda, will probably start looking at adding that this week. i've gotten as far as nodes trying to tftp their respective token-$id file with errors. i imagine it would also be required for debugging deploy issues in gate | 06:56 |
| *** pradipta is now known as pradipta_away | 07:09 | |
| *** pradipta_away is now known as pradipta | 07:09 | |
| *** mrda is now known as mrda_away | 07:11 | |
| *** mdurnosvistov_lt has joined #openstack-ironic | 07:15 | |
| *** lazy_prince has quit IRC | 07:24 | |
| *** killer_p- has joined #openstack-ironic | 07:24 | |
| *** eghobo has quit IRC | 07:28 | |
| *** mdurnosvistov_lt has quit IRC | 07:31 | |
| *** vkozhukalov has quit IRC | 07:38 | |
| openstackgerrit | Adam Gandelman proposed a change to openstack/python-ironicclient: Avoid traceback with insufficient auth credentials https://review.openstack.org/79494 | 07:42 |
| *** romcheg1 has joined #openstack-ironic | 07:43 | |
| *** rsacharya has joined #openstack-ironic | 07:44 | |
| GheRivero | morning all | 07:47 |
| romcheg1 | Morning GheRivero | 07:59 |
| *** ifarkas has joined #openstack-ironic | 08:05 | |
| *** sanek11 is now known as agordeev | 08:11 | |
| *** jistr has joined #openstack-ironic | 08:13 | |
| *** jistr has quit IRC | 08:20 | |
| *** matsuhashi has quit IRC | 08:21 | |
| *** jistr has joined #openstack-ironic | 08:22 | |
| openstackgerrit | Fengqian Gao proposed a change to openstack/ironic: Keep python 3.X compatibility for xrange https://review.openstack.org/79875 | 08:39 |
| *** yuriyz has joined #openstack-ironic | 08:41 | |
| *** saju_m has quit IRC | 08:41 | |
| *** max_lobur has joined #openstack-ironic | 08:44 | |
| *** ekarlso has quit IRC | 08:47 | |
| *** ekarlso has joined #openstack-ironic | 08:47 | |
| *** max_lobur1 has joined #openstack-ironic | 08:54 | |
| *** max_lobur has quit IRC | 08:57 | |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Improve Ironic Conductor threading & locks https://review.openstack.org/78515 | 08:58 |
| *** athomas has joined #openstack-ironic | 08:58 | |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Improve Ironic Conductor threading & locks https://review.openstack.org/78515 | 09:03 |
| mdurnosvistov | Morning Ironic!!! :) | 09:05 |
| *** saju_m has joined #openstack-ironic | 09:09 | |
| *** derekh has joined #openstack-ironic | 09:12 | |
| *** max_lobur1 has quit IRC | 09:13 | |
| *** max_lobur has joined #openstack-ironic | 09:13 | |
| *** jrist has quit IRC | 09:27 | |
| *** ifarkas has quit IRC | 09:28 | |
| *** romcheg1 has quit IRC | 09:31 | |
| *** ifarkas has joined #openstack-ironic | 09:31 | |
| *** rsacharya has quit IRC | 09:34 | |
| *** killer_prince has quit IRC | 09:34 | |
| *** killer_p- is now known as killer_prince | 09:34 | |
| *** jrist has joined #openstack-ironic | 09:40 | |
| *** romcheg1 has joined #openstack-ironic | 09:41 | |
| *** romcheg has left #openstack-ironic | 09:43 | |
| *** romcheg has quit IRC | 09:43 | |
| *** romcheg1 is now known as romcheg | 09:43 | |
| *** athomas has quit IRC | 09:43 | |
| *** tatyana has joined #openstack-ironic | 09:43 | |
| *** lucasagomes has joined #openstack-ironic | 09:47 | |
| *** lucasagomes has quit IRC | 09:48 | |
| *** lucasagomes has joined #openstack-ironic | 09:48 | |
| *** killer_prince2 has joined #openstack-ironic | 09:50 | |
| *** athomas has joined #openstack-ironic | 09:51 | |
| openstackgerrit | Ghe Rivero proposed a change to openstack/ironic: Set boot device to PXE when deploying https://review.openstack.org/79892 | 10:07 |
| *** lucasagomes has quit IRC | 10:10 | |
| *** lucasagomes has joined #openstack-ironic | 10:10 | |
| *** martyntaylor has joined #openstack-ironic | 10:18 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: PXE validate() to fail if no Ironic API URL https://review.openstack.org/79579 | 10:43 |
| *** rsacharya has joined #openstack-ironic | 10:44 | |
| *** killer_prince2 has quit IRC | 10:45 | |
| *** saju_m has quit IRC | 10:49 | |
| *** killer_prince2 has joined #openstack-ironic | 10:53 | |
| *** rsacharya has quit IRC | 11:01 | |
| openstackgerrit | Ghe Rivero proposed a change to openstack/ironic: Update SeaMicro to use MixinVendorInterface https://review.openstack.org/79903 | 11:04 |
| *** bigjools has quit IRC | 11:06 | |
| *** bigjools has joined #openstack-ironic | 11:07 | |
| *** bigjools has joined #openstack-ironic | 11:07 | |
| *** bigjools has quit IRC | 11:11 | |
| *** bigjools has joined #openstack-ironic | 11:12 | |
| *** bigjools has joined #openstack-ironic | 11:12 | |
| *** dtantsur has quit IRC | 11:19 | |
| *** ChanServ changes topic to "test/gate jobs are queuing now in preparation for gerrit maintenance at 12:00 utc (eta to resume is 12:30 utc)" | 11:27 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Make the Nova Ironic driver to wait for ACTIVE https://review.openstack.org/79906 | 11:34 |
| *** loki_ has quit IRC | 11:59 | |
| *** linggao has joined #openstack-ironic | 12:05 | |
| *** ndipanov_gone is now known as ndipanov | 12:08 | |
| *** dtantsur has joined #openstack-ironic | 12:19 | |
| *** ChanServ changes topic to "gerrit on review.openstack.org is down for maintenance (revised eta to resume is 13:00 utc)" | 12:24 | |
| *** lucasagomes is now known as lucas-hungry | 12:36 | |
| lsmola | lucas-hungry: hello | 12:37 |
| *** saju_m has joined #openstack-ironic | 12:37 | |
| *** openstackgerrit has quit IRC | 12:54 | |
| *** openstackgerrit has joined #openstack-ironic | 12:54 | |
| *** jbjohnso_ has joined #openstack-ironic | 12:58 | |
| *** openstackgerrit has quit IRC | 13:08 | |
| *** openstackgerrit has joined #openstack-ironic | 13:08 | |
| *** pradipta is now known as pradipta_away | 13:20 | |
| *** rloo has joined #openstack-ironic | 13:24 | |
| *** jdob has joined #openstack-ironic | 13:26 | |
| *** lucas-hungry is now known as lucasagomes | 13:37 | |
| lucasagomes | lsmola, hey there | 13:37 |
| lsmola | lucasagomes: hello | 13:39 |
| lsmola | lucasagomes: I just wanted to ask and the exact match of flavors with nodes | 13:39 |
| lsmola | lucasagomes: with nova baremetal, it doesn't do the exact match for me | 13:39 |
| lsmola | lucasagomes: are you using anything special for Ironic? | 13:40 |
| lsmola | lucasagomes: this functionality should be done with nova filters, right? | 13:40 |
| lucasagomes | lsmola, hmm unless something changed recently in nova bm it should | 13:40 |
| lucasagomes | lsmola, we use a diff scheduler for nova bm and ironic | 13:40 |
| *** jgrimm has quit IRC | 13:40 | |
| lsmola | lucasagomes: ha | 13:41 |
| lsmola | lucasagomes: and nova bm, uses also different scheduler than normal nova ? | 13:42 |
| lucasagomes | lsmola, for ironic scheduler_host_manager = ironic.nova.scheduler.ironic_host_manager.IronicHostManager | 13:42 |
| lucasagomes | lsmola, yes | 13:42 |
| lucasagomes | it should | 13:42 |
| lucasagomes | because baremetal can't do partial usage of a machine | 13:42 |
| lucasagomes | u always use 100% | 13:42 |
| *** rloo has quit IRC | 13:42 | |
| lsmola | lucasagomes: right | 13:42 |
| lsmola | lucasagomes: hm I saw the managers code | 13:43 |
| lsmola | lucasagomes: but this seems to be oriented for eating of the resources | 13:43 |
| *** nosnos has quit IRC | 13:43 | |
| lsmola | lucasagomes: though i don't see a logic for picking the right resources | 13:43 |
| *** dkehn_ has joined #openstack-ironic | 13:43 | |
| lsmola | lucasagomes: that should be still done by nova filters, right? | 13:44 |
| lsmola | lucasagomes: or am I missing something? :-) | 13:44 |
| lucasagomes | right, lemme take a look at nova bm | 13:45 |
| lsmola | lucasagomes: ok, thanks | 13:46 |
| lsmola | lucasagomes: the nova bm manager looked almost the same for me | 13:46 |
| lsmola | lucasagomes: so seems like with other manager, this is still in place http://docs.openstack.org/developer/nova/devref/filter_scheduler.html | 13:47 |
| lucasagomes | yeah the scheduler_driver still using the FilterScheduler | 13:48 |
| lucasagomes | which calls the host_manager driver to make a decision | 13:48 |
| *** matsuhashi has joined #openstack-ironic | 13:50 | |
| lsmola | lucasagomes: ok so we might need to write some new filters | 13:52 |
| *** matty_dubs|gone is now known as matty_dubs | 13:52 | |
| lsmola | lucasagomes: or at least by quick peek, I haven't found a great ones for the exact match | 13:52 |
| lucasagomes | right yeah I have to take a quick look into it as well, as we leave the scheduling job for nova I'm not very familiar in the area | 13:53 |
| lucasagomes | I'll take a look into it | 13:53 |
| *** matsuhashi has quit IRC | 13:56 | |
| lucasagomes | lsmola, ur correct! | 13:58 |
| lucasagomes | lsmola, did a quick test here, created a node with more resources than the flavor | 13:58 |
| lucasagomes | nova pick it anyway | 13:58 |
| * lucasagomes files a bug | 13:58 | |
| lucasagomes | devananda, ^ | 13:59 |
| lsmola | lucasagomes: right, so for icehouse scope, we can't probably use this | 13:59 |
| lsmola | lucasagomes: haven't had time to dig deep into existing filters | 14:00 |
| *** matsuhashi has joined #openstack-ironic | 14:00 | |
| lsmola | lucasagomes: ok, this is worth a bug in tripleo nova element | 14:00 |
| agordeev | morning all! | 14:00 |
| lucasagomes | lsmola, yeah, I'm filling one for ironic right now | 14:00 |
| lsmola | lucasagomes: we need to list a proper filters in nova.conf | 14:00 |
| lucasagomes | lsmola, http://paste.openstack.org/show/73250/ | 14:01 |
| lucasagomes | lsmola, yeah I will take a look into the filters as well | 14:01 |
| lsmola | lucasagomes: ok, I will create a TripleO bug | 14:02 |
| lucasagomes | ack thanks | 14:02 |
| agordeev | could anyone know whether v2.0 ending is really needed for admin_url for ironic.nova.driver? is it okay? | 14:02 |
| lucasagomes | agordeev, admin_url? | 14:03 |
| lucasagomes | or auth_uri? | 14:03 |
| *** openstack has joined #openstack-ironic | 14:04 | |
| lucasagomes | agordeev, ah, I think it's the same case | 14:04 |
| lucasagomes | it seems to be mandatory | 14:04 |
| lucasagomes | since keystone clients don't work with versionless urls | 14:05 |
| lucasagomes | agordeev, https://bugs.launchpad.net/ironic/+bug/1290482 | 14:05 |
| *** fungi has joined #openstack-ironic | 14:07 | |
| *** ChanServ sets mode: +o fungi | 14:07 | |
| *** fungi changes topic to "OpenStack Bare Metal Provisioning | Docs: http://docs.openstack.org/developer/ironic/ | Bugs: https://bugs.launchpad.net/ironic | Status: https://etherpad.openstack.org/p/IronicWhiteBoard" | 14:07 | |
| *** ChanServ sets mode: -o fungi | 14:07 | |
| *** fungi has left #openstack-ironic | 14:07 | |
| agordeev | lucasagomes: same thing will happen if admin_url from ironic nova driver is versionless too. | 14:07 |
| NobodyCam | Good Morning Ironic | 14:07 |
| agordeev | NobodyCam: morning | 14:07 |
| NobodyCam | morning agordeev :) | 14:08 |
| lucasagomes | agordeev, right yeah, mind commenting that on the bug? | 14:08 |
| lucasagomes | NobodyCam, ^ the scheduler thing | 14:08 |
| lucasagomes | NobodyCam, nova is not doing an exact matching when picking a node | 14:08 |
| NobodyCam | humm | 14:09 |
| lucasagomes | NobodyCam, should we have a filter for the nova scheduler as well? | 14:11 |
| *** rloo has joined #openstack-ironic | 14:11 | |
| lucasagomes | NobodyCam, or only the host_manager would be able to prevent a node from being pick by nova if it doesn't match the exact same resources? | 14:11 |
| devananda | morning, all | 14:11 |
| NobodyCam | lucasagomes: I thought we did >= matching | 14:12 |
| NobodyCam | morning devananda :) | 14:12 |
| lucasagomes | NobodyCam, no | 14:12 |
| agordeev | devananda: morning | 14:12 |
| lucasagomes | we should cause we can't do pick part of the resources | 14:12 |
| lsmola | lucasagomes: could you link your bug with this? https://bugs.launchpad.net/tripleo/+bug/1291396 | 14:12 |
| lucasagomes | also we would be giving the client a machine that has more resources than he asked and is paying for | 14:12 |
| lucasagomes | devananda, morning | 14:13 |
| lucasagomes | devananda, ^ | 14:13 |
| lucasagomes | lsmola, will do | 14:13 |
| lucasagomes | lsmola, thanks | 14:13 |
| NobodyCam | lucasagomes: ya the host manager will comsume the entire node | 14:13 |
| lucasagomes | NobodyCam, so not really | 14:13 |
| devananda | lucasagomes: just read scroll back ... ya, that needs to be fixed | 14:13 |
| lucasagomes | I gave it a go here | 14:13 |
| lucasagomes | it substract the amout of resources specified in the node but the n-cpu still showing the difference as available | 14:14 |
| *** rloo_ has joined #openstack-ironic | 14:14 | |
| *** rloo has quit IRC | 14:14 | |
| lucasagomes | devananda, right | 14:14 |
| lucasagomes | devananda, and the way to fix it would be writing an filter? | 14:14 |
| lucasagomes | a filter* | 14:14 |
| devananda | lucasagomes: ya | 14:14 |
| lucasagomes | devananda, ack | 14:15 |
| lucasagomes | lsmola, ^ | 14:15 |
| *** rloo_ has quit IRC | 14:15 | |
| *** rloo has joined #openstack-ironic | 14:16 | |
| lsmola | devananda: right, one or more filter, right now they have separate filter for each metric, e.g. CPUFIlter | 14:16 |
| devananda | yea.. | 14:16 |
| devananda | urgh | 14:16 |
| devananda | i was just looking at it | 14:16 |
| devananda | lsmola: i suspect it's goign to be more than one filter difference | 14:16 |
| lsmola | derekh: https://github.com/openstack/nova/tree/master/nova/scheduler/filters | 14:16 |
| lsmola | derekh: sorry, it was for devananda | 14:17 |
| devananda | lsmola: do you see any reason that we'd need to change filter_scheduler.py itself? or just filters/* | 14:17 |
| lsmola | devananda: though it make sense to just do baremetal_filter | 14:17 |
| lsmola | devananda: we should be fine with just filters I guess | 14:18 |
| derekh | lsmola: np | 14:18 |
| lsmola | devananda: cause we will need to use other filters .e.g for affinity, etc. | 14:19 |
| lsmola | devananda: in the future | 14:19 |
| devananda | ya | 14:19 |
| lsmola | devananda: though it might be that much of them will be baremetal specific, using some Ironic metadata | 14:20 |
| *** matsuhashi has quit IRC | 14:23 | |
| devananda | lucasagomes: in our ironic_host_manager, we should probably start setting a different list of default filters | 14:24 |
| devananda | https://github.com/openstack/nova/blob/master/nova/scheduler/host_manager.py#L45 | 14:24 |
| lucasagomes | devananda, will take a look, thanks | 14:25 |
| lucasagomes | devananda, NobodyCam btw https://review.openstack.org/#/c/79906/ | 14:25 |
| NobodyCam | lucasagomes: awesome :) | 14:26 |
| devananda | lucasagomes: good stuff! one issue, comment added | 14:27 |
| lucasagomes | devananda, ack | 14:28 |
| devananda | oh, and another | 14:28 |
| * devananda makes coffee | 14:29 | |
| lucasagomes | devananda, ah indeed, that doesn't looks at the deleted instances etc | 14:29 |
| lucasagomes | devananda, will fix it | 14:29 |
| *** matsuhashi has joined #openstack-ironic | 14:30 | |
| NobodyCam | I lucasagomes would we want a time out from nova, Ie if networking is messed up and node is unable to ping back, we'll be stuck in waiting for call back? | 14:33 |
| NobodyCam | s/I // | 14:35 |
| lucasagomes | NobodyCam, yeah I thought about it | 14:35 |
| lucasagomes | NobodyCam, but hmm it's hard to say the timeout unless we set it to something high | 14:35 |
| lucasagomes | ironic has a timeout so it would set to state to deployfail | 14:36 |
| lucasagomes | so I was relying on that for nova | 14:36 |
| * NobodyCam notes that he has not had much coffee, but say data networking is messed up, but ipmi net is working | 14:37 | |
| NobodyCam | the node could poweron but then never ping to start deploy | 14:38 |
| NobodyCam | ya it would have to be a big timeout | 14:39 |
| devananda | NobodyCam: in that case, ironic will eventually time it out | 14:39 |
| devananda | nova-bm had a pxe_deploy_timeout option which defaulted (badly) to 0 | 14:40 |
| devananda | so I suspect nova is not going to impose any other timeouts here | 14:40 |
| NobodyCam | ahh so our default should be like 5 minutes? | 14:43 |
| NobodyCam | for wait for call back to deploying change? | 14:43 |
| *** jgrimm has joined #openstack-ironic | 14:44 | |
| devananda | NobodyCam: http://git.openstack.org/cgit/openstack/ironic/tree/ironic/conductor/manager.py#n91 | 14:44 |
| NobodyCam | ahh :) | 14:45 |
| * NobodyCam goes back to coffee | 14:45 | |
| devananda | GheRivero: ping re: https://review.openstack.org/#/c/71332/ | 14:50 |
| GheRivero | morning all | 14:51 |
| GheRivero | devananda: pong: https://review.openstack.org/#/c/79892/ | 14:52 |
| *** killer_prince2 is now known as lazy_prince | 14:52 | |
| GheRivero | I though the patch was going to be much more different, but didn't realize until late | 14:52 |
| devananda | GheRivero: now that the multiplexing has landed, I'd like to get that one in too | 14:53 |
| devananda | GheRivero: are you going to work on it? if not, i will take a look | 14:54 |
| GheRivero | yes. it-s redone in the one I previously mentioned | 14:54 |
| GheRivero | https://review.openstack.org/#/c/79892/ | 14:54 |
| devananda | oh! right, i see :) | 14:54 |
| devananda | GheRivero: I see a comment about the SSH driver, but it looks like it will raise an unhandled exception | 14:58 |
| devananda | GheRivero: you validate here https://review.openstack.org/#/c/79892/1/ironic/conductor/utils.py but for the pxe_ssh driver, that's going to raise InvalidParameterValue, no? | 14:59 |
| GheRivero | yeah | 14:59 |
| GheRivero | i should catch that | 14:59 |
| devananda | :) | 14:59 |
| *** saju_m has quit IRC | 15:03 | |
| *** saju_m has joined #openstack-ironic | 15:04 | |
| *** lazy_prince2 has joined #openstack-ironic | 15:04 | |
| jroll | morning y'all | 15:06 |
| NobodyCam | morning jroll :) | 15:06 |
| devananda | mornin | 15:06 |
| *** lazy_prince has quit IRC | 15:07 | |
| lucasagomes | morning jroll | 15:10 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Driver utils should raise unsupported method https://review.openstack.org/79964 | 15:10 |
| devananda | GheRivero: you may want to rebase your fix on ^ so you catch just the right error | 15:11 |
| GheRivero | thx. will do | 15:11 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Make the Nova Ironic driver to wait for ACTIVE https://review.openstack.org/79906 | 15:12 |
| *** matsuhashi has quit IRC | 15:18 | |
| devananda | lucasagomes: there's an interesting race in the nova driver | 15:19 |
| *** lazy_prince2 is now known as lazy_prince | 15:19 | |
| devananda | lucasagomes: not that your patch is creating it, but i see it now | 15:19 |
| devananda | lucasagomes: if an instance is deleted from within ironic during provisioning | 15:20 |
| devananda | lucasagomes: ^^ will abort the spawn() call with an exception, leaving teh instance in an error state | 15:20 |
| *** dwalleck_ has joined #openstack-ironic | 15:21 | |
| devananda | lucasagomes: and a subsequent call to nova destroy() will fail because | 15:21 |
| devananda | node_uuid = self._require_node(instance) | 15:21 |
| devananda | will also raise notfound | 15:21 |
| *** dwalleck has joined #openstack-ironic | 15:22 | |
| devananda | I suspect it's not possible to trigger those circumstances purely via the nova api -- i think someone would have to issue "ironic node-update delete instance_uuid" to cause it | 15:22 |
| lucasagomes | ohhhh | 15:22 |
| devananda | but there are lots of odd paths through nova code that i dont know | 15:22 |
| lucasagomes | well we can't delete the instance if the node is provisioned I think | 15:23 |
| lucasagomes | or being provisioned, lemme check | 15:23 |
| agordeev | jroll: morning! I'm very excited about teeth-agent :) | 15:23 |
| devananda | right, true | 15:23 |
| lucasagomes | devananda, yeah there's many corner cases | 15:23 |
| jroll | agordeev: :) | 15:23 |
| devananda | lucasagomes: yep :( | 15:23 |
| lucasagomes | it's hard to figure out all them, we need some intensive tests creating and deleting instances | 15:23 |
| lucasagomes | messing around to detect | 15:23 |
| devananda | lucasagomes: the checks around "dont do X if the node is associated" are specifically to prevent this sort of edge case | 15:24 |
| lucasagomes | devananda, it's fine :) it's part of the maturity of the project | 15:24 |
| devananda | lucasagomes: let's hope they work as intended! | 15:24 |
| lucasagomes | devananda, yes | 15:24 |
| devananda | just thought I'd point this one out as I saw it related to your patch :) | 15:24 |
| devananda | lucasagomes: which, btw, looks good. just one more question i think | 15:24 |
| devananda | lucasagomes: should we be raising any more specific exceptions? | 15:24 |
| *** dwalleck_ has quit IRC | 15:25 | |
| lucasagomes | devananda, for the nova driver? hmm it might be important yes, lemme check other drivers to see what they r raising | 15:25 |
| lucasagomes | I've been using NovaException with custom messages all the way | 15:26 |
| lucasagomes | :( | 15:26 |
| lucasagomes | shame on me | 15:26 |
| devananda | hehe | 15:26 |
| *** martyntaylor has left #openstack-ironic | 15:27 | |
| devananda | lucasagomes: http://paste.openstack.org/show/73262/ | 15:27 |
| devananda | not that the nova-bm driver is always right, but that gives some starting point | 15:27 |
| lucasagomes | devananda, def | 15:28 |
| lucasagomes | i should be using InstanceDeployFailure for my patch | 15:28 |
| lucasagomes | will change it | 15:28 |
| lucasagomes | the starting point yay! | 15:28 |
| lucasagomes | and InstanceNotFound as well | 15:28 |
| devananda | :) | 15:30 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Make the Nova Ironic driver to wait for ACTIVE https://review.openstack.org/79906 | 15:39 |
| *** saju_m has quit IRC | 15:40 | |
| Shrews | adam_g: can you send me your entire localrc you're using? | 15:43 |
| *** bashok has quit IRC | 15:46 | |
| openstackgerrit | A change was merged to openstack/ironic: Improve Ironic Conductor threading & locks https://review.openstack.org/78515 | 15:47 |
| lucasagomes | dear lazy#ironic: what tool do you guys use to create diagrams in linux? | 15:49 |
| agordeev | Shrews: adam_g: hey guys. Just want to thank you for working on making devstack patch better! :) Let me know if i need to add you as co-author to commit message. Or feel free to add this information by yourself :) | 15:49 |
| Shrews | agordeev: np. i still can't get the vm's to boot, but i think this may be a problem with my setup | 15:50 |
| devananda | lucasagomes: dia || openoffice | 15:51 |
| agordeev | lucasagomes: if online services count, lastly i've used draw.io | 15:51 |
| lucasagomes | devananda, right yeah, I started something on the openoffice but it's kinda painful | 15:51 |
| lucasagomes | thanks guys :) will take a look | 15:51 |
| devananda | agordeev: have you gotten VMs to finish provisioning and be accessible via ssh? | 15:51 |
| devananda | lucasagomes: yea... it is ... | 15:51 |
| *** romcheg has quit IRC | 15:52 | |
| agordeev | devananda: in my local env that happens flawlessly | 15:53 |
| *** lazy_prince has quit IRC | 15:55 | |
| agordeev | devananda: but I can't login to them using ssh key from nova | 15:55 |
| *** matty_dubs is now known as matty_dubs|lunch | 15:56 | |
| devananda | agordeev: hm, so what is different in your env? many of us are not getting a POST back from the ramdisk to ironic-api | 15:56 |
| *** dwalleck has quit IRC | 15:56 | |
| devananda | Shrews: are you still hitting ^ ? or did you guys fix that over night? | 15:56 |
| *** eghobo has joined #openstack-ironic | 15:56 | |
| Shrews | devananda: yeah, i'm still stuck there even using adam_g's new ip settings | 15:57 |
| Shrews | thus my request to him for his localrc | 15:57 |
| *** eghobo has quit IRC | 15:58 | |
| agordeev | devananda: sounds like ironic-api isn't accessible from baremetal instance's network. Have you checked that connectivity? | 15:58 |
| devananda | agordeev: right. can you share your localrc? | 15:59 |
| agordeev | devananda: regarding my env, i just run ./stack.sh on my desktop. Nothing special :) | 15:59 |
| devananda | definitely something with the networking, either neutron or the bridge | 15:59 |
| devananda | agordeev: not in a VM, but on your desktop directly? | 15:59 |
| agordeev | devananda: right, on desktop directly. Makes sense? | 16:00 |
| devananda | agordeev: the VMs you are creating -- are they running in bridge mode or NAT or .. ? | 16:00 |
| devananda | agordeev: we are testing within cloud instances in HPCS and RackSpace -- this is where the gate will run, so this is unltimately where it has to work | 16:01 |
| agordeev | devananda: let me restart stack.sh, i'll check what mode they're using | 16:01 |
| agordeev | devananda: http://paste.openstack.org/show/73265/ FYI, my localrc. | 16:03 |
| *** eghobo has joined #openstack-ironic | 16:04 | |
| Shrews | agordeev: you should be using nova master branch now | 16:05 |
| agordeev | Shrews: right, but it works with old branch too :) | 16:10 |
| agordeev | devananda: network mode. It uses libvirt network. | 16:11 |
| agordeev | devananda: are you able to run arbitrary shell commands in that cloud instances? | 16:14 |
| *** jistr has quit IRC | 16:14 | |
| *** digambar has joined #openstack-ironic | 16:16 | |
| *** vkozhukalov has joined #openstack-ironic | 16:18 | |
| *** lazy_prince has joined #openstack-ironic | 16:18 | |
| *** lazy_prince has quit IRC | 16:19 | |
| agordeev | devananda: any chance for me to get the same cloud image for test? i'll test it against a VM then | 16:22 |
| *** romcheg has joined #openstack-ironic | 16:22 | |
| NobodyCam | devananda: question on one of lifeless's comments: re 72696, its about line 87 of https://review.openstack.org/#/c/72969/8/scripts/devtest_undercloud.sh | 16:27 |
| agordeev | devananda: also it's possible to switch network for regular linux bridge, not ovs. Additionaly this may make libvirt upgrade probably unnecessary. | 16:29 |
| Shrews | agordeev: i can create you an hp vm instance temporarily. i'll need your public key | 16:31 |
| openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Default value for database.connection config option https://review.openstack.org/79787 | 16:34 |
| devananda | rloo: we shouldn't make changes directly to anything in ironic.openstack.common -- these are synced from oslo | 16:36 |
| rloo | devananda. then it isn't really possible to set a default value for connection. | 16:37 |
| devananda | rloo: seems like this is a bug in oslo, then | 16:38 |
| rloo | let me forward you email from Roman (one of the people that made the change) | 16:38 |
| devananda | k | 16:39 |
| rloo | devananda: you're at devananda.vdv@gmail.com? | 16:39 |
| devananda | ya | 16:39 |
| devananda | rloo: your change looks fine to me, except that i think it should be applid here https://github.com/openstack/oslo-incubator/blob/master/openstack/common/db/options.py | 16:40 |
| rloo | devananda. it *was* there and it was removed: https://review.openstack.org/#/c/74737/ | 16:41 |
| devananda | ... i see | 16:41 |
| devananda | dhellmann: around? | 16:41 |
| *** matty_dubs|lunch is now known as matty_dubs | 16:42 | |
| devananda | rloo: see this? https://github.com/openstack/nova/blob/master/nova/config.py#L29 | 16:42 |
| devananda | perhaps we should do something similar | 16:42 |
| rloo | devananda. I tried that. but it didn't show up when generating the sample. maybe i need to learn more about how to use that generate sample script. | 16:43 |
| devananda | rloo: it's OK if the sample config line is empty | 16:44 |
| rloo | devananda. OH, in that case... | 16:44 |
| rloo | devananda. although I'm not convinced it is really OK. but if you're Ok with it ;) | 16:44 |
| rloo | devananda. Are you OK if we don't set the connection to .../$sqlite_db? cuz that also broke in oslo. | 16:45 |
| rloo | devananda. (cuz in oslo sqlite_db was moved to 'database' from DEFAULT. and the templating only works if you refer to variables from DEFAULT. We can just set it to ironic.sqlite. | 16:46 |
| devananda | rloo: hrmm | 16:47 |
| *** eghobo has quit IRC | 16:47 | |
| rloo | devananda. Which btw, someone modified our copy of openstack/common/db/options.py so it is set to ironic.sqlite. Which is why I thought it was OK to change that file. | 16:48 |
| *** eghobo has joined #openstack-ironic | 16:48 | |
| devananda | rloo: so the issue is more that the defaults do not work right now, than that the config file line is blank | 16:56 |
| devananda | i'm surprised theyre not the same in nova | 16:56 |
| NobodyCam | devananda: have a quick second to help me digest lifeless's comment? | 16:56 |
| devananda | sure | 16:56 |
| rloo | devananda: I'm looking now at nova, to see why they seem to be able to get the config default in their sample. their generator sample is diff from ours and I'm not actually familiar with that code. | 16:57 |
| NobodyCam | 09:27 | NobodyCam > devananda: question on one of lifeless's comments: re 72696, its about line 87 of https://review.openstack.org/#/c/72969/8/scripts/devtest_undercloud.sh | 16:57 |
| devananda | rloo: our generate_sample comes from oslo. theirs may not? | 16:57 |
| rloo | devananda: dunno yet. or they may be diff versions from oslo. | 16:58 |
| rloo | devananda. wrt the defaults -- it works for them cuz they aren't using $sqlite_db in their value anymore. We could do that too, just set connection default value = .../ironic.sqlite instead of $sqlite_db. | 16:59 |
| devananda | rloo: it sounds like a change to https://github.com/openstack/ironic/blob/master/ironic/common/config.py is the simplest solution,a nd replicates what Nova is doing | 17:00 |
| rloo | devananda: ok, so in nova, their sample has connection=<None>. So they have the same problem we have. | 17:00 |
| devananda | rloo: ack. i'm fine with that | 17:01 |
| devananda | rloo: it's not ideal, but it's better than carrying a significant local change to openstack.common | 17:02 |
| devananda | NobodyCam: ok, i think i undersatnd the comments there. what is your question(s)? | 17:02 |
| rloo | devananda. ok then. but i think we need to change https://github.com/openstack/ironic/blob/master/ironic/common/service.py, not config.py (config.py only used by tests) | 17:02 |
| rloo | devananda. thx. | 17:02 |
| devananda | rloo: ooh. well | 17:03 |
| devananda | rloo: nvp. that should be fine. I was about to say ironic-dbsync doesn't use service.py, but it does :) | 17:03 |
| devananda | *nvm | 17:03 |
| *** max_lobur has quit IRC | 17:04 | |
| NobodyCam | devananda: its the compile vs end point I not sure about | 17:04 |
| *** harlowja_away is now known as harlowja | 17:04 | |
| rloo | sorry, devananda. so are you OK with this modified since it is from oslo: https://github.com/openstack/ironic/blob/master/ironic/openstack/common/db/options.py | 17:04 |
| NobodyCam | s/I/I'm. | 17:04 |
| rloo | devananda. I mean it was modified to have sqlite_db default = ironic.sqlite (instead of oslo.sqlite i think) | 17:04 |
| devananda | rloo: that's a normal change as part of the sync process :) | 17:06 |
| devananda | rloo: https://github.com/openstack/oslo-incubator/blob/master/update.py | 17:06 |
| devananda | rloo: syncing oslo-incubator to a project does a bunch of string replacement | 17:07 |
| *** yuriyz has quit IRC | 17:07 | |
| *** rsacharya has joined #openstack-ironic | 17:07 | |
| rloo | devananda. thx. so much to learn... What happens when the stuff is out of oslo.incubation? Or I'll wait to see what happens... | 17:08 |
| comstud | devananda: did you still need something? | 17:08 |
| *** rpodolyaka has joined #openstack-ironic | 17:09 | |
| rpodolyaka | morning ironic | 17:09 |
| devananda | NobodyCam: erm, I *think* what lifeless means is that we should have a unified heat template that works for both nova-bm and ironic | 17:09 |
| devananda | SpamapS: thoughts on ^ ? I don't know heat enough to know how feasible taht is | 17:10 |
| openstackgerrit | Roman Podoliaka proposed a change to openstack/ironic: Set the database.connection option default value https://review.openstack.org/80003 | 17:10 |
| devananda | comstud: hi! paging in that question ... :) | 17:11 |
| NobodyCam | 09:27 | NobodyCam SpamapS question is on lifeless's comments: re 72696, about line 87 of https://review.openstack.org/#/c/72969/8/scripts/devtest_undercloud.sh | 17:11 |
| rpodolyaka | rloo: ^ this is how I'd like to see the problem with the default value of database.connection option not set be solved for ironic-dbsync calls | 17:11 |
| NobodyCam | oh bad paste | 17:11 |
| devananda | comstud: how does nova handle passing information about failed requests across RPC ? | 17:11 |
| NobodyCam | oh lucasagomes still around? | 17:11 |
| lucasagomes | NobodyCam, hey yes | 17:12 |
| devananda | comstud: right now, ironic is raising exceptions from the conductor, which propagate across RPC to the API, where they are converted to clean user-facing error messages and returned with the right HTTP status code | 17:12 |
| * SpamapS arouses upon seeing highlight | 17:12 | |
| rloo | rpodolyaka: you mean the discussion with deva, not NobodyCam's link? | 17:12 |
| SpamapS | devananda: unified for both .. sort of :) | 17:12 |
| devananda | comstud: but this results in a lot of exception traces in the conductor log for "normal" errors | 17:12 |
| rpodolyaka | rloo: no, https://review.openstack.org/80003 :) | 17:13 |
| rloo | rpodolyaka: sorry, i see your link now. | 17:13 |
| SpamapS | devananda: I don't think we can really do "if you chose this parameter, set all of these keys, otherwise those" at parse/deploy time... | 17:13 |
| NobodyCam | lucasagomes: what do yo think about a change like this? http://paste.openstack.org/show/uc8D59LQozRNOXMNzUDR/ | 17:13 |
| devananda | SpamapS: thus the need for a different heat template? | 17:13 |
| SpamapS | devananda: what we can do is have the things that are common to nova-bm and ironic in one template, and then the things that are different in another.. and build appropriately. | 17:13 |
| devananda | SpamapS: right | 17:14 |
| rpodolyaka | rloo: the reason for this is that we're not going to restore the default value in oslo.db (as it makes no sense when oslo.db becomes a separate library) | 17:14 |
| devananda | SpamapS: it appears that lifeless is requesting a single unified template in the comment which NobodyCam linked above | 17:14 |
| SpamapS | devananda: note that the new OS::Heat::SoftwareConfig resource may allow us to do it at deploy time. | 17:14 |
| devananda | SpamapS: but i may have misunderstood his comment | 17:14 |
| NobodyCam | SpamapS: I'm just unclear in what lifeless is wanting in that comment | 17:14 |
| * SpamapS reads | 17:14 | |
| rloo | rpodolyaka: thx. did you test those changes are are those suggestions? cuz config.py is only used by ironic tests. | 17:15 |
| lucasagomes | NobodyCam, nice, to get the return value from that LoopingCall | 17:15 |
| lucasagomes | NobodyCam, looks good :0 | 17:15 |
| NobodyCam | yea!! | 17:15 |
| lucasagomes | :)* | 17:15 |
| lucasagomes | +1 | 17:15 |
| lifeless | devananda: I'm suggesting like we do for overcloud.yaml, to parameterise the build via variables rather than a different target name. This might be a bad idea :) | 17:16 |
| rpodolyaka | rloo: hmm, I expected to be used when the service starts | 17:16 |
| lifeless | devananda: but I'm looking down the track at how we maintain all the different fragments | 17:16 |
| * NobodyCam thanks jroll for the point in the right direction | 17:16 | |
| *** saju_m has joined #openstack-ironic | 17:16 | |
| rpodolyaka | rloo: nevertheless, now the default value is set in db api module - and it's used in both tests and runtime | 17:16 |
| NobodyCam | morning lifeless :) | 17:17 |
| rpodolyaka | rloo: (and ironic-dbsync) | 17:17 |
| comstud | devananda: There's a decorator you can put on rpc methods to not log certain expected exceptions | 17:17 |
| rloo | rpodolyaka: actually, i think your changes work. and anyway, even though tests uses config.py, i seem to recall that tests overrode the connection setting anyway ;) | 17:17 |
| devananda | lifeless: ideally, i agree that a single undercloud yaml which can handle the array of [virt | hardware] x [nova-bm | ironic] would be great | 17:17 |
| devananda | comstud: oooh, perfect | 17:17 |
| comstud | devananda: The decorator is buried in the oslo messaging module somewhere | 17:17 |
| comstud | lemme see if I can find a nvoa example for you | 17:18 |
| rpodolyaka | rloo: yeah, they'd better do - we don't want to write sqlite files, when we can write to memory much faster :) | 17:18 |
| devananda | comstud: thanks | 17:18 |
| devananda | rloo: all tests in the gate use mysql or pgsql, AIUI | 17:18 |
| comstud | devananda: @messaging.expected_exceptions(NotImplementedError) | 17:18 |
| *** tatyana has quit IRC | 17:18 | |
| comstud | messaging = oslo.messaging | 17:18 |
| rpodolyaka | devananda: rloo: yes, but only for migrations tests, as far as I recall | 17:19 |
| devananda | rpodolyaka: not for unit tests? | 17:19 |
| rloo | rpodolyaka: should we just use your review then? I don't mind if you take over. the only thing (that i can think of now) is that the default sql connection is wrong, in api.py. | 17:19 |
| rpodolyaka | devananda: not yet, I think | 17:19 |
| devananda | hrm | 17:19 |
| adam_g | Shrews, http://paste.ubuntu.com/7080336/ | 17:20 |
| rpodolyaka | devenanda: we've done some work in oslo.db to make it easier to run tests on real backends, but it's still WIP. maybe in Juno | 17:20 |
| NobodyCam | lifeless: do you see that type of overcloud refactor as part of this patch (72969)? | 17:21 |
| devananda | comstud: it looks like we're still using oslo-incubator/rpc | 17:21 |
| rpodolyaka | rloo: the only reason why I uploaded the change in the first place, is that it's easier to show the code right away :) | 17:21 |
| *** digambar has quit IRC | 17:21 | |
| rpodolyaka | rloo: hmm, what would you put there instead? | 17:22 |
| comstud | devananda: it's in there too somewhere | 17:22 |
| comstud | probably with the same name | 17:22 |
| *** digambar has joined #openstack-ironic | 17:22 | |
| comstud | we were using this before oslo.messaging | 17:22 |
| devananda | not exactly the same name :) | 17:22 |
| devananda | will keep searching | 17:22 |
| comstud | haha | 17:22 |
| comstud | k | 17:22 |
| rloo | rpodolyaka: i actually prefer what you have, but it isn't the same default t value that it used to be. | 17:23 |
| comstud | devananda: 428 def client_exceptions(*exceptions): | 17:23 |
| comstud | (line # may be different..t his was from an old branch) | 17:24 |
| comstud | rpc/common.py | 17:24 |
| rloo | rpodolyaka: connection=sqlite:////ironic/openstack/common/db/$sqlite_db | 17:24 |
| rpodolyaka | rloo: oh, I see what you are talking about. I'd argue that it's better to put sqlite database within the state path :) | 17:24 |
| devananda | comstud: yep! looks like it. testing now | 17:24 |
| adam_g | Shrews, are you s till having tftp connectivity issues? | 17:24 |
| rpodolyaka | rloo: (even if it has a default value now) | 17:25 |
| rloo | rpodolyaka: i'd argue the same thing as you. | 17:25 |
| rloo | devananda: sorry, are you OK if we set the connection default value to something different that what it was set to before? | 17:25 |
| devananda | rloo: such as? | 17:25 |
| rloo | something like sqlite:////ironic/ironic.sqlite | 17:26 |
| rloo | instead of ironic/openstack/common/db. | 17:26 |
| devananda | rloo: so functionally the same, but literaly different | 17:26 |
| devananda | sure | 17:26 |
| rloo | right. also means breaks people's dev environments. but better now than later? | 17:26 |
| devananda | yep | 17:26 |
| rloo | good. | 17:26 |
| devananda | def note that change in teh commit msg | 17:26 |
| devananda | and pls check the docs for any reference -- i dont rememver one, but good to heck | 17:27 |
| devananda | check | 17:27 |
| rpodolyaka | ok | 17:27 |
| rloo | rpodolyaka: ^^ let's just use your review. Can you take care of what deva asks or do you want me to? | 17:27 |
| rpodolyaka | rloo: I'll take care of it, no problem | 17:28 |
| rloo | thx rpodolyaka! | 17:28 |
| rpodolyaka | rloo: np | 17:28 |
| *** rsacharya has quit IRC | 17:31 | |
| *** openstackstatus has quit IRC | 17:38 | |
| *** openstackstatus has joined #openstack-ironic | 17:38 | |
| NobodyCam | devananda: new rev pushed up to https://review.openstack.org/#/c/72969 | 17:39 |
| SpamapS | NobodyCam: ok, NOW I'm reading the comments.. my browser crashed before.. -P | 17:39 |
| NobodyCam | lol | 17:39 |
| NobodyCam | SpamapS: I just clobbered it | 17:39 |
| NobodyCam | but may happy to address it still | 17:40 |
| NobodyCam | s/may/sm/ | 17:40 |
| NobodyCam | s/sm/am/ | 17:40 |
| NobodyCam | :-p | 17:40 |
| * NobodyCam refills coffee cup | 17:40 | |
| SpamapS | NobodyCam: ok so what lifeless was referring to is that Heat can substitute things based on "environment".. but I don't think that is applicable here. | 17:40 |
| lifeless | devananda: I'm not saying a single undercloud.yaml; I'm saying a single make target that builds the appropriate undercloud.yaml | 17:41 |
| SpamapS | Because AFAIK, environments only allow substitution at the "type" level. | 17:41 |
| openstackgerrit | Roman Podoliaka proposed a change to openstack/ironic: Set the database.connection option default value https://review.openstack.org/80003 | 17:41 |
| SpamapS | lifeless: right, I think that is entirely doable. | 17:41 |
| SpamapS | though I feel our makefile command lines are getting almost criminal in length | 17:42 |
| openstackgerrit | Yuriy Zveryanskyy proposed a change to openstack/ironic: Update Neutron if mac address of the port changed https://review.openstack.org/77939 | 17:44 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Suppress conductor logging of expected exceptions https://review.openstack.org/80022 | 17:44 |
| devananda | ^ breaks unit tests - anyone want to fix it up? | 17:45 |
| devananda | lifeless: ah. ok. | 17:45 |
| devananda | gotta jump on a conf call ... bbiah | 17:45 |
| *** jbjohnso_ has quit IRC | 17:50 | |
| *** rpodolyaka has quit IRC | 17:52 | |
| *** rpodolyaka has joined #openstack-ironic | 17:56 | |
| lucasagomes | devananda, quick q, about the delete state, should we allow a "nova delete" while the instance is being deployed? | 17:56 |
| lucasagomes | cause ironic prevents that (node will be locked) | 17:56 |
| devananda | lucasagomes: during deploywat - yes. at other times ... yes... but not sure how | 17:56 |
| lucasagomes | devananda, ok | 17:56 |
| lucasagomes | devananda, one wait to do this is to make destroy() keep retrying | 17:57 |
| lucasagomes | one way* | 17:57 |
| lucasagomes | keep retrying sending the destroy() command to ironic | 17:57 |
| *** rpodolyaka has quit IRC | 17:57 | |
| *** athomas has quit IRC | 17:57 | |
| lucasagomes | although it means that the deploy will be complete and then it will be destroyed | 17:57 |
| lucasagomes | it's not going to abort it | 17:57 |
| devananda | lucasagomes: yea. interrupt in-progress tasks .... | 17:58 |
| devananda | in juno | 17:58 |
| lucasagomes | ack | 17:58 |
| devananda | lucasagomes: for now, we want nova to not be broken when a delete can't delete | 17:58 |
| lucasagomes | devananda, there's anything u can return to nova saying | 17:58 |
| devananda | some sanity around the retry would be good, with a FIXME | 17:58 |
| lucasagomes | "look i can't delete it now" | 17:58 |
| *** vkozhukalov has quit IRC | 17:58 | |
| devananda | lucasagomes: i would look for 503 NODE LOCKED | 17:58 |
| lucasagomes | devananda, right, yeah we already do it | 17:59 |
| lucasagomes | problem is if it's timeout | 17:59 |
| devananda | and as long as the node has the same instance uuid, keep retrying | 17:59 |
| lucasagomes | so it works, but the timeout right now is pretty small | 17:59 |
| devananda | which timeout? | 17:59 |
| lucasagomes | I think it's 5 times 2 seconds | 17:59 |
| lucasagomes | that's the default time out | 17:59 |
| lucasagomes | or 10 times /me checks | 17:59 |
| lucasagomes | 5 times with an interval of 2 seconds each | 18:00 |
| lucasagomes | that's what the "_retry_if_service_is_unavailable" uses | 18:00 |
| *** rpodolyaka has joined #openstack-ironic | 18:01 | |
| lucasagomes | (it's configurable via conf options tho) | 18:01 |
| lucasagomes | anyhoo, I will give it some thought | 18:01 |
| lucasagomes | devananda, please Chris Behens comment at line 455 https://review.openstack.org/#/c/79906/3/ironic/nova/virt/ironic/driver.py | 18:06 |
| lucasagomes | I think we don't need that check/block there, as u pointed out. If the node is deleted it's going to raise InstanceNotFound(after I update the patch to replace the NodeNotFound) | 18:06 |
| lucasagomes | so nova will handle that | 18:07 |
| *** eghobo has quit IRC | 18:07 | |
| *** derekh has quit IRC | 18:07 | |
| *** jbjohnso_ has joined #openstack-ironic | 18:09 | |
| *** eghobo has joined #openstack-ironic | 18:09 | |
| NobodyCam | oh poop... forgot to set USE_IRONIC .... doh | 18:18 |
| lucasagomes | gotta run | 18:18 |
| *** lucasagomes is now known as lucas-afk | 18:18 | |
| *** rpodolyaka has quit IRC | 18:20 | |
| *** jistr has joined #openstack-ironic | 18:22 | |
| *** rpodolyaka has joined #openstack-ironic | 18:23 | |
| *** sandeepr has joined #openstack-ironic | 18:26 | |
| *** sandeepr has left #openstack-ironic | 18:26 | |
| *** killer_prince is now known as lazy_prince | 18:26 | |
| Shrews | adam_g: my vm boots still fail. not sure of the cause. how do i verify that it's a tftp issue? | 18:28 |
| adam_g | Shrews, i connect to the system via libvirt/virt-manager and watch the console | 18:28 |
| adam_g | Shrews, alternatively you can bump the verbosity of atftp and log to a file to see what the node is grabbing | 18:28 |
| adam_g | i found that the ramdisk is attempting to grab their token-$uuid files from the wrong tftp path, https://bugs.launchpad.net/diskimage-builder/+bug/1291533 | 18:29 |
| *** saju_m has quit IRC | 18:30 | |
| adam_g | testing now with a fedora ramdisk to see if that was ever expected to work | 18:30 |
| *** tatyana has joined #openstack-ironic | 18:39 | |
| *** digambar has quit IRC | 18:41 | |
| *** jistr has quit IRC | 18:44 | |
| *** rpodolyaka has quit IRC | 18:47 | |
| *** jbjohnso_ has quit IRC | 18:52 | |
| *** jb11211 has joined #openstack-ironic | 18:58 | |
| *** tatyana has quit IRC | 18:59 | |
| devananda | back | 19:00 |
| NobodyCam | wb devananda | 19:01 |
| devananda | lucas-afk: ahh, i rhink you're right | 19:01 |
| *** tatyana has joined #openstack-ironic | 19:01 | |
| *** jbjohnso_ has joined #openstack-ironic | 19:04 | |
| *** rpodolyaka has joined #openstack-ironic | 19:05 | |
| comstud | lucasagomes: let me know if you have any questions about my review | 19:07 |
| comstud | lucasagomes: nova will set instance to ERROR on any exception.. except for things like InstanceNotFound... which it treats as a race with delete, etc | 19:08 |
| comstud | non-exception return will set ACTIVE | 19:08 |
| comstud | lucas-afk: ^ | 19:09 |
| devananda | comment added re: ironic_states.DELETED | 19:13 |
| *** ndipanov is now known as ndipanov_gone | 19:13 | |
| comstud | ok | 19:18 |
| comstud | devananda: left a reply to your comment! | 19:20 |
| comstud | bbiab | 19:20 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Set boot device to PXE when deploying https://review.openstack.org/79892 | 19:30 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Driver utils should raise unsupported method https://review.openstack.org/79964 | 19:30 |
| *** rpodolyaka has quit IRC | 19:31 | |
| *** rpodolyaka has joined #openstack-ironic | 19:31 | |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Set boot device to PXE when deploying https://review.openstack.org/79892 | 19:31 |
| NobodyCam | devananda: almost :-p http://paste.openstack.org/show/y28vi0IJLl5r8TLNLrRk/ | 19:32 |
| devananda | needs reviews: https://review.openstack.org/#/c/79892/ and https://review.openstack.org/#/c/79964/2 | 19:33 |
| devananda | needs soemone to fix unit tests: https://review.openstack.org/#/c/80022/1 | 19:33 |
| devananda | also needs unit tests: https://review.openstack.org/#/c/79192/ | 19:35 |
| devananda | also needs another core review: https://review.openstack.org/#/c/79191/ | 19:39 |
| Shrews | devananda: i'll pick up 80022 for you | 19:40 |
| devananda | Shrews: awesome, ty | 19:40 |
| devananda | rloo: if you have tiem to review/vote on https://review.openstack.org/#/c/80003/, that'd be great. at a glance, it looks good to me | 19:41 |
| rloo | devananda. sure, I forgot about that ;) Will look at it now. | 19:41 |
| Shrews | omg... provisioning_state = active | 19:42 |
| devananda | Shrews: !!! \o/ | 19:42 |
| Shrews | devananda: thank adam_g | 19:42 |
| NobodyCam | w00t !!! | 19:42 |
| adam_g | Shrews, that tftp path issue was the last blocker for you? | 19:43 |
| Shrews | yes! | 19:43 |
| adam_g | cool | 19:43 |
| Shrews | and i can successfully ssh in | 19:43 |
| adam_g | :) | 19:43 |
| devananda | i need to run out for appointments, will be on and off for the remainder of the day | 19:44 |
| adam_g | gonna start replacing atftpd usage in devstack with tftpd-hpa to match tripleo-incubator's undercloud | 19:44 |
| Shrews | so, the symlink fix i suggested works | 19:44 |
| adam_g | agordeev, ^ | 19:44 |
| devananda | will respond to msgs here as i can | 19:44 |
| Shrews | not sure if that's the correct solution or not | 19:44 |
| Shrews | adam_g: that works too. probably the better solution, though more involved | 19:46 |
| *** Lotus907efi has joined #openstack-ironic | 19:54 | |
| *** rpodolyaka has quit IRC | 20:20 | |
| *** rpodolyaka has joined #openstack-ironic | 20:21 | |
| *** rpodolyaka1 has joined #openstack-ironic | 20:21 | |
| *** mrda_away is now known as mrda | 20:21 | |
| *** Slower_ is now known as Slower | 20:22 | |
| *** MarkAtwood has joined #openstack-ironic | 20:41 | |
| *** linggao has quit IRC | 20:54 | |
| NobodyCam | rloo: just read your comment on set_boot_device and seamicro | 20:58 |
| NobodyCam | that seems like a bug in SM to me | 20:59 |
| rloo | NobodyCam: i don't know if that is a(nother) bug or not. wanted to point it out but it shouldn't stop the patch. | 20:59 |
| NobodyCam | ya | 20:59 |
| rloo | NobodyCam. I think it was just a docstring thing that I -1'd. can't remember now. I probably should be more explicit about what I am really concerned about ;) | 21:00 |
| *** vkozhukalov has joined #openstack-ironic | 21:02 | |
| NobodyCam | oh brb | 21:02 |
| *** rpodolyaka1 has quit IRC | 21:05 | |
| *** matty_dubs is now known as matty_dubs|gone | 21:05 | |
| *** rpodolyaka has joined #openstack-ironic | 21:09 | |
| Shrews | devananda: aarrgh. you are the devil. 80022 takes some finagling to verify the real exception | 21:13 |
| *** jdob has quit IRC | 21:13 | |
| *** jbjohnso_ has quit IRC | 21:20 | |
| *** romcheg has quit IRC | 21:31 | |
| *** rpodolyaka has quit IRC | 21:52 | |
| *** rpodolyaka has joined #openstack-ironic | 21:56 | |
| *** rpodolyaka has quit IRC | 22:03 | |
| NobodyCam | greghaynes: ping | 22:10 |
| *** rpodolyaka has joined #openstack-ironic | 22:10 | |
| greghaynes | Hey | 22:10 |
| NobodyCam | hey hey was just looking at the tr -d of the newlines comment | 22:11 |
| NobodyCam | https://review.openstack.org/#/c/72969/13/scripts/devtest_undercloud.sh | 22:11 |
| greghaynes | Yep | 22:12 |
| NobodyCam | wanted to ask did you like the remove as the tr did or would escaping (ie. s/\n/\\n/) be more what you had in mind | 22:13 |
| greghaynes | I dont really like either - I think ideally we pass in the key as-is | 22:13 |
| greghaynes | that was my fix for not being competent enough in bash quote escaping to allow the newlines to be passed | 22:13 |
| greghaynes | oh, maybe thats what you mean by escaping? | 22:15 |
| greghaynes | if so then yes to that :) | 22:15 |
| NobodyCam | let me try a test here | 22:17 |
| openstackgerrit | A change was merged to openstack/ironic: Mark hash_replicas as experimental https://review.openstack.org/79191 | 22:22 |
| *** jgrimm has quit IRC | 22:26 | |
| *** rpodolyaka has quit IRC | 22:28 | |
| *** rpodolyaka has joined #openstack-ironic | 22:30 | |
| *** max_lobur has joined #openstack-ironic | 22:30 | |
| *** harlowja has quit IRC | 22:30 | |
| *** harlowja has joined #openstack-ironic | 22:31 | |
| *** max_lobur has quit IRC | 22:38 | |
| *** max_lobur has joined #openstack-ironic | 22:39 | |
| greghaynes | NobodyCam: haha, thats an interesting choice ;) | 22:44 |
| NobodyCam | greghaynes: ok pushed up another rev. testing with Ironic | 22:44 |
| NobodyCam | lol | 22:44 |
| NobodyCam | its json encoded | 22:44 |
| greghaynes | yes, which doesnt deal with newlines to it has to escape :p | 22:44 |
| *** MarkAtwood has quit IRC | 22:45 | |
| NobodyCam | oh :-p | 22:46 |
| greghaynes | should work though so *shrug* | 22:46 |
| * greghaynes will try and give it a proper test in a few | 22:46 | |
| NobodyCam | greghaynes: TY :) | 22:47 |
| devananda | Shrews: I'll take that as a compliment :) | 22:47 |
| NobodyCam | wb devananda :) | 22:47 |
| adam_g | Shrews, agordeev devananda : just pushed another patchset to https://review.openstack.org/#/c/70348/ -- swaps atftpd out in favor of tftpd-hpa (to align with tripleO) and fixes the issues we've been hitting around fetching auth token via tftp. | 22:47 |
| adam_g | i'm getting functional ironic nodes with working metadata now on a fresh devstack run | 22:47 |
| NobodyCam | greghaynes: but it will still need the grep I sucked out | 22:50 |
| greghaynes | Really? I added that grep... I think ideally we can just send the data in the original format | 22:52 |
| NobodyCam | I was looking at a echo and saw two newlines at the end | 22:53 |
| NobodyCam | figured that should have only been one | 22:53 |
| *** MarkAtwood has joined #openstack-ironic | 22:53 | |
| greghaynes | ah. That may be within the margin of error for things which use the key | 22:53 |
| NobodyCam | oh ya, no that grep is not what we need. it removes the --begin ... --end lines | 22:55 |
| *** eguz has joined #openstack-ironic | 22:56 | |
| greghaynes | yerp | 22:59 |
| *** eghobo has quit IRC | 23:00 | |
| *** eguz has quit IRC | 23:00 | |
| *** tatyana has quit IRC | 23:06 | |
| devananda | adam_g: \o/ | 23:07 |
| * devananda tries | 23:07 | |
| *** tatyana has joined #openstack-ironic | 23:11 | |
| *** lsmola has quit IRC | 23:36 | |
| *** lsmola has joined #openstack-ironic | 23:36 | |
| devananda | adam_g: have you guys been rebasing on devstack trunk? | 23:37 |
| adam_g | devananda, good point--no, i haven't | 23:37 |
| adam_g | tho ive been checking out master and cherry-picking from that review | 23:38 |
| adam_g | without a conflict in a while | 23:38 |
| devananda | cool | 23:39 |
| devananda | i'm taking a slightly old local VM that I have, updating everything from git,a nd trying devstack in offline mode | 23:39 |
| devananda | risky, yes, but HPCS is not functioning fo rme right now | 23:39 |
| devananda | and this cafe wifi is slow | 23:39 |
| adam_g | offline mode eh? | 23:40 |
| adam_g | its going to need to hit the internet to build the ramdisk, in case you dont have that stuff cached | 23:40 |
| *** lucas-afk has quit IRC | 23:46 | |
| devananda | i probably do | 23:47 |
| devananda | but | 23:47 |
| devananda | it would be neat if devstack knew how to pass the offline flag to DIB | 23:47 |
| devananda | I gotta run again in a bit, but if anyone's looking for patches to fix up -- https://review.openstack.org/#/c/79964/ | 23:50 |
| *** tatyana has quit IRC | 23:54 | |
| *** jbjohnso_ has joined #openstack-ironic | 23:54 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!