Wednesday, 2019-01-09

« Tuesday, 2019-01-08 Index Thursday, 2019-01-10 »
*** rcernin has quit IRC00:07
*** rcernin has joined #openstack-ironic00:09
*** etingof has quit IRC00:42
*** sthussey has quit IRC00:51
*** jesusaur has joined #openstack-ironic01:12
*** tiendc has joined #openstack-ironic01:20
openstackgerritJulia Kreger proposed openstack/ironic master: WIP ipv6 job  https://review.openstack.org/61316901:35
*** dustinc has quit IRC01:40
openstackgerritinspurericzhang proposed openstack/ironic-tempest-plugin master: Update hacking version to latest  https://review.openstack.org/62934702:14
openstackgerritJulia Kreger proposed openstack/ironic master: WIP ipv6 job  https://review.openstack.org/61316902:48
*** rloo has quit IRC02:51
*** Qianbiao has joined #openstack-ironic03:00
*** gyee has quit IRC03:10
openstackgerritQianBiao Ng proposed openstack/ironic-specs master: Add spec for Huawei iBMC driver  https://review.openstack.org/62872403:35
openstackgerritMerged openstack/sushy-tools master: Add configuration file to sushy-emulator  https://review.openstack.org/61617803:48
*** dustinc has joined #openstack-ironic04:11
*** dustinc has quit IRC04:22
*** dustinc has joined #openstack-ironic04:24
*** dustinc has quit IRC04:35
*** dustinc has joined #openstack-ironic04:36
*** dustinc has quit IRC04:43
*** arne_wiebalck_ has joined #openstack-ironic04:53
*** hwoarang has joined #openstack-ironic04:56
*** lifeless_ has joined #openstack-ironic04:57
*** baha has joined #openstack-ironic05:01
*** dnuka has joined #openstack-ironic05:02
*** jesusaur has quit IRC05:02
*** hwoarang_ has quit IRC05:02
*** irclogbot_1 has quit IRC05:02
*** trungnv has quit IRC05:02
*** arne_wiebalck has quit IRC05:02
*** chason has quit IRC05:02
*** lifeless has quit IRC05:02
*** arne_wiebalck_ is now known as arne_wiebalck05:02
dnukagood morning o/05:02
dnukagood morning arne_wiebalck :)05:02
*** irclogbot_1 has joined #openstack-ironic05:04
*** jesusaur has joined #openstack-ironic05:09
*** dustinc has joined #openstack-ironic05:20
*** dustinc has quit IRC05:29
openstackgerritJulia Kreger proposed openstack/ironic master: WIP ipv6 job  https://review.openstack.org/61316905:34
openstackgerritKaifeng Wang proposed openstack/ironic master: Expose conductors: api-ref  https://review.openstack.org/62286206:16
*** e0ne has joined #openstack-ironic06:22
*** e0ne has quit IRC06:23
*** pcaruana has joined #openstack-ironic07:06
*** moshele has joined #openstack-ironic07:10
*** etingof has joined #openstack-ironic07:17
*** e0ne has joined #openstack-ironic07:18
*** e0ne has quit IRC07:22
arne_wiebalckdnuka: good morning!07:23
dnukao/07:23
*** rcernin has quit IRC07:26
*** hjensas has quit IRC07:31
*** e0ne has joined #openstack-ironic07:47
*** rpittau has joined #openstack-ironic07:54
rpittaugood morning ironic! o/07:54
dnukagood morning rpittau :)07:54
rpittauhi dnuka :)07:54
iurygregorygood morning everyone o/08:35
rpittauhey iurygregory :)08:35
dnukamorning iurygregory08:35
*** e0ne has quit IRC08:47
* iurygregory really love the TimeoutException =( 08:51
*** S4ren has joined #openstack-ironic09:00
*** dougsz has joined #openstack-ironic09:13
*** hjensas has joined #openstack-ironic09:13
* etingof feels that being in hurry and running openstack do not always play well... ;)09:23
etingofgood morning everybody o/09:23
dnukamorning etingof09:23
etingofdnuka, so did `tox -r` help?09:24
iurygregorymorning etingof o/09:24
dnukaetingof, yes :)09:24
etingofawesome09:24
rpittauhi etingof :)09:24
*** moshele has quit IRC09:24
dnukafolks, I cannot push changes to gerrit, "Problem running 'git remote update gerrit'"09:26
iurygregorycan you check your .gitreview file in the repo?09:27
etingofis it what `git review` ends up with?09:27
dnukaack09:27
dnukaProblem running 'git remote update gerrit'09:28
dnukaFetching gerrit09:28
dnukadnuka@review.openstack.org: Permission denied (publickey).09:28
dnukafatal: Could not read from remote repository.09:28
iurygregorywell you key seem to have wrong permisions O.o09:28
dnukawell, everything worked yestreday :)09:29
dnuka*yesterday09:29
etingofcan it be that ssh key pass phrase has expired in the ssh agent?09:29
* etingof would run `ssh -v dnuka@review.openstack.org` to see where auth fails09:30
dnukaack09:30
dnukaPermission denied :(09:31
iurygregory a workaround woulb be build a new key and add to gerrit =)09:32
etingofso `git config -l` should tell you the exact ssh command which git uses to access remote repo09:32
etingofthen dnuka should make sure that that ssh command (including -p <port> perhaps) works out alone09:33
dnukaetingof, iurygregory got it :)09:33
etingofalong the way ssh should tell dnuka which ssh key(s) it is trying to use and if they get rejected09:33
dnukaack09:34
*** robbbe has joined #openstack-ironic09:36
etingofif dnuka suspects his local repo is somehow misconfigured, he could clone the upstream repo somewhere (e.g. in /tmp) and (if cloning works out) see if `git config -l` differs...09:36
*** moshele has joined #openstack-ironic09:37
dnukaetingof, got it09:37
*** derekh has joined #openstack-ironic09:38
dnukaetingof, cloning works! and git config -l` output differs09:41
etingofdnuka, the configuration is stored in .git/config so you can experiment with these files09:42
dnukaack09:42
*** zhangfei has joined #openstack-ironic09:43
etingofdnuka, just make sure not to accidentally lose the work you've done but not yet pushed upstream09:43
iurygregoryat least what is in gerrit is not lost =D09:43
dnukaetingof, almost forgot that :)09:44
iurygregorygit review -d <309:44
dnukaiurygregory, yes :)09:44
* rpittau looks at the clock, it's coffee time!09:48
dnuka:)09:49
iurygregorycoffee \o/09:49
* etingof can't get his clock reading in the morning unless initial coffeination is over09:52
dnuka:D09:52
*** S4ren has quit IRC09:53
rpittaulol09:57
*** dnuka is now known as dnuka|brb10:02
*** dtantsur|afk is now known as dtantsur10:10
dtantsurmorning ironic10:11
*** iurygregory has quit IRC10:12
*** dnuka has joined #openstack-ironic10:13
dnukamorning dtantsur10:16
*** S4ren has joined #openstack-ironic10:18
*** e0ne has joined #openstack-ironic10:21
rpittauhi dtantsur :)10:21
*** iurygregory has joined #openstack-ironic10:26
*** stendulker has joined #openstack-ironic10:27
*** zhangfei has quit IRC10:32
openstackgerritDhanuka Warusadura proposed openstack/sushy master: [WIP] Introduce ``convert_to_date``  https://review.openstack.org/62588910:32
openstackgerritDhanuka Warusadura proposed openstack/sushy master: [WIP] Introduce ``convert_to_date``  https://review.openstack.org/62588910:36
*** iurygregory has quit IRC10:50
*** moshele has quit IRC10:57
*** stendulker has quit IRC11:11
*** iurygregory has joined #openstack-ironic11:16
*** dnuka has quit IRC11:16
*** dnuka has joined #openstack-ironic11:16
iurygregorydtantsur, morning o/11:23
*** hjensas has quit IRC11:24
iurygregoryits normal a node have deploy_interface as None?11:26
openstackgerritHamdy Khader proposed openstack/python-ironicclient master: Add 'hostname' to port's local link connection  https://review.openstack.org/62877311:26
rpittauiurygregory, is that on ironic conf ?11:32
iurygregoryrpittau, nope when looking at the node info11:32
rpittauI think it can be empty11:33
iurygregorywell according to our guide we should set the node in maintenance but we cant lol11:34
*** moshele has joined #openstack-ironic11:34
iurygregoryopenstack baremetal node maintenance set d209a91d-5bea-4901-9809-a06697859494 --reason "Changing driver and/or hardware interfaces"11:34
iurygregoryNo valid host was found. Reason: No conductor service registered which supports driver pxe_ipmitool for conductor group "". (HTTP 400)11:34
dtantsuriurygregory: this is expected. no updates are possible with an invalid driver.11:35
dtantsuriurygregory: also deploy_interface=None is expected for classic drivers11:35
dtantsuralso we may be debugging the same thing :)11:36
iurygregorycool11:36
iurygregoryyep we are =P11:36
* iurygregory not sure if you saw my in email in the thread lol11:36
dtantsurit's an actual bug in ironic :(11:37
iurygregory=(11:39
iurygregoryso our doc is wrong when we say that we can put the node in maintenance =(11:40
iurygregory'It is not recommended to change the interfaces for active nodes. If absolutely needed, the nodes have to be put in the maintenance mode first'11:40
dtantsuriurygregory: no, this is correct.11:40
dtantsurit's not a workaround for updating broken nodes though11:41
iurygregoryoh ok11:41
openstackgerritDmitry Tantsur proposed openstack/ironic master: Fix updating nodes with removed or broken drivers  https://review.openstack.org/62944711:45
dtantsuriurygregory: ^^^11:45
dtantsurTheJulia, mgoddard if you're around ^^^11:46
iurygregorydtantsur, cool you rock =D11:46
mgoddarddtantsur: I'll take a look11:47
openstackgerritHamdy Khader proposed openstack/python-ironicclient master: Add is-smartnic port attribute to port command  https://review.openstack.org/62944911:51
openstackgerritDerek Higgins proposed openstack/ironic master: [WIP] Replace use of Q_USE_PROVIDERNET_FOR_PUBLIC  https://review.openstack.org/62114611:59
*** iurygregory has quit IRC12:00
*** dnuka has quit IRC12:06
*** hjensas has joined #openstack-ironic12:07
openstackgerritDmitry Tantsur proposed openstack/ironic stable/rocky: Fix updating nodes with removed or broken drivers  https://review.openstack.org/62945112:10
*** dnuka has joined #openstack-ironic12:13
*** iurygregory has joined #openstack-ironic12:15
*** rpittau is now known as rpittau|lunch12:20
openstackgerritDmitry Tantsur proposed openstack/ironic master: Allocation API: allow picking random conductor for RPC topic  https://review.openstack.org/62925012:23
openstackgerritDmitry Tantsur proposed openstack/ironic master: Allocation API: conductor API (without HA and take over)  https://review.openstack.org/62822312:23
*** tiendc has quit IRC12:27
*** e0ne has quit IRC12:30
*** dnuka has quit IRC12:37
openstackgerritIury Gregory Melo Ferreira proposed openstack/networking-baremetal master: Change networking-baremetal to zuulv3/python3  https://review.openstack.org/61402612:39
openstackgerritIury Gregory Melo Ferreira proposed openstack/ironic-inspector master: Follow-up Zuulv3 (Ironic jobs rename)  https://review.openstack.org/62947513:05
*** trown|outtypewww is now known as trown13:09
*** jcoufal has joined #openstack-ironic13:16
*** rpittau|lunch is now known as rpittau13:21
*** e0ne has joined #openstack-ironic13:24
*** vkmc has left #openstack-ironic13:29
*** Qianbiao has quit IRC13:51
openstackgerritIury Gregory Melo Ferreira proposed openstack/ironic master: Remove dsvm from zuulv3 jobs  https://review.openstack.org/62917313:54
*** rloo has joined #openstack-ironic13:55
*** moshele has quit IRC13:57
openstackgerritIury Gregory Melo Ferreira proposed openstack/ironic-inspector master: Follow-up Zuulv3 (Ironic jobs rename)  https://review.openstack.org/62947513:58
openstackgerritIury Gregory Melo Ferreira proposed openstack/ironic-tempest-plugin master: Rename zuulv3 jobs  https://review.openstack.org/62949414:07
*** baha has joined #openstack-ironic14:16
openstackgerritIlya Etingof proposed openstack/sushy master: Add System<->Manager linkage follow up  https://review.openstack.org/62950014:19
openstackgerritDmitry Tantsur proposed openstack/ironic master: Allocation API: conductor API (without HA and take over)  https://review.openstack.org/62822314:26
TheJuliadtantsur: do we have an elastic recheck in place for the failed to read errors?14:32
dtantsurTheJulia: nope. at least not from me. I can look into it.14:35
dtantsur(and thanks for approving that bug fix)14:35
TheJuliadtantsur: thanks, I wonder just how often we're seeing it14:35
openstackgerritIury Gregory Melo Ferreira proposed openstack/networking-generic-switch master: Rename ironic job to match zuulv3  https://review.openstack.org/62950814:35
iurygregorymorning TheJulia o/14:36
dtantsurTheJulia: quite often I'm sure14:36
dtantsurI wonder if the recent IPA change is involved.. or is it just bionic14:36
dtantsurTheJulia: do you have an example elastic recheck update handy?14:36
TheJuliadtantsur: i don't sadly, but I am super curious about that change14:37
rpittaugood morning TheJulia :)14:37
TheJuliaI never really agreed with it, tbh14:37
TheJuliabut I wasn't going to block it14:37
dtantsurwhy didn't you agree?14:38
TheJuliabecause we shouldn't make assumptions into the contents we write14:41
TheJuliaHow do we know what is being written has or needs a partition table?14:41
TheJuliashouldn't be an issue on any of the platforms really used with ironic though, for most common things14:41
openstackgerritDmitry Tantsur proposed openstack/ironic master: [WIP] Allocation API: implementation  https://review.openstack.org/62926014:42
dtantsurTheJulia: hmm.. maybe you're right..14:43
dtantsurTheJulia: as to elastic-recheck, I cannot even make logstash output me *anything* >_<14:43
TheJuliadtantsur: ugh :(14:43
TheJuliamaybe clarkb has some magical insight there he can point us to?14:44
TheJuliare platform stuffs, I am quite altruistic though, not quite practical in that regard, so it is likely practical, but worth digging at that change. I can dig at that with a fresh set of eyes once coffee hits and I'm done with email14:45
dtantsurokay, the syntax is funny, but I seem to start figuring it out14:46
dtantsurTheJulia: 95 hits in a week for message:"Failed to create config drive on disk" AND message:"test -e" :(14:50
TheJuliaEEK14:50
TheJuliaWe had a lot of failures late last week, so that could be it14:50
TheJuliatest -e seems... a little not specific enough because I think it is body contents matching not line level matching14:51
* TheJulia has no idea how the cluster is indexed14:51
dtantsurTheJulia: well, it seems to find what we need14:52
dtantsurnow creating a story for this14:52
*** baha has quit IRC14:52
TheJulia\o/14:53
TheJuliaalmost done with email, so I can begin deep diving on the failure14:54
dtantsurTheJulia: https://storyboard.openstack.org/#!/story/200474414:54
TheJuliaawesome14:55
iurygregoryyay finally finished the patchs to rename jobs \o/14:58
dtantsurTheJulia: the query shows that at least sometimes it failed on xenial, so we may be able to rule out bionic as a cause15:04
dtantsurTheJulia: and happens only with wholedisk+direct jobs.. which make the IPA patch an even bigger suspect15:05
TheJuliahmmm15:06
dtantsurhmm, wait, the IPA code runs for all image types15:06
TheJuliaWell, it is clear the person with huawei is awake right now, they've responded to two emails so far this morning15:07
*** chason_ has joined #openstack-ironic15:09
*** mjturek has joined #openstack-ironic15:10
dtantsurTheJulia: hmm, does this actually mean that we should run partprobe (again!) http://logs.openstack.org/47/629447/1/check/ironic-tempest-dsvm-ipa-wholedisk-direct-tinyipa-multinode/65739ae/compute1/logs/ironic-bm-logs/subnode-0_no_ansi_2019-01-09-12:38:08_log.txt.gz#_2019-01-09_12_37_27_550 ?15:12
TheJuliaThat is likely what is happening, but that would only fix the symptom15:13
* TheJulia wonders15:13
dtantsurand why for love's sake are we using GPT in a BIOS job? Oo15:13
iurygregorymagic15:14
TheJuliaoO15:14
dtantsurbecause yes, we do: http://logs.openstack.org/47/629447/1/check/ironic-tempest-dsvm-ipa-wholedisk-direct-tinyipa-multinode/65739ae/compute1/logs/ironic-bm-logs/subnode-0_no_ansi_2019-01-09-12:38:08_log.txt.gz#_2019-01-09_12_37_21_81515:14
*** baha has joined #openstack-ironic15:14
TheJuliadtantsur: the image we are writing ?15:15
dtantsurI mean, it is supported, but why are we doing it?15:15
dtantsurwell, yes, that's the explanation15:15
TheJuliawe recently bumped the version of the image, I suspect it is a silent change15:16
dtantsurwe recently moved to new version of cirros, can it bring GPT here? /cc derekh15:16
TheJuliathat is what I'm thinking15:16
TheJuliaI'm glad we put in the gpt repair logic15:17
dtantsurvda: vda1 vda1515:17
dtantsurWUT15:17
dtantsurno, srsly WUT15:17
dtantsurdear linux..15:17
derekhdtantsur: dunno15:18
TheJuliawut^wutwut15:19
iurygregoryO.o15:19
dtantsurI need whiskey. a lot of whiskey.15:20
iurygregoryafter see this i need to agree15:20
TheJulia2 is the config drive, 15 is the /boot volume15:21
* TheJulia suspects it is too early to open the japanese whiskey15:22
dtantsuryeah, except that, I suspect, the configdrive is now always 2...15:22
dtantsurI have some great stuff from Islay from my birthday. But too early even for me..15:22
* iurygregory miss Japanese beer.15:22
dtantsuranother guess is that we need to insert partprobe into yet another place15:23
dtantsurthis is a better option for us15:23
dtantsurwe can even try going down this path first15:24
TheJulialooks like it would have to be a whdi to hit this failure15:25
TheJuliadtantsur: ++15:26
TheJulialooks like we should just be able to do it in ipa which would cover that case... feels like we have another bug someplace in the path15:29
dtantsurwho knows... I'm patching ironic-lib for now15:30
dtantsurTheJulia: I wonder if https://review.openstack.org/#/c/594922/ may be involved15:33
patchbotpatch 594922 - ironic-lib - Check GPT table with sgdisk insread of partprobe (MERGED) - 2 patch sets15:33
TheJuliaSo we pulled down the partition image...15:33
openstackgerritIlya Etingof proposed openstack/sushy master: Add Chassis<->ComputerSystem/Manager linkage  https://review.openstack.org/62302815:33
TheJuliaerr no15:34
TheJuliayeah, wholedisk is what it was15:34
iurygregory"nat" is correct?15:35
openstackgerritDmitry Tantsur proposed openstack/ironic-lib master: [WIP] Run partprobe after adding a configdrive partition  https://review.openstack.org/62954715:36
dtantsurTheJulia: let's see what the CI says bout ^^^15:36
TheJulia++15:37
TheJuliacirros is not starting out with a partition table :\15:40
iurygregoryTheJulia, would be necessary push a change to openstack-infra/project-config since the have the name for one job that is changed? http://codesearch.openstack.org/?q=ironic-tempest-dsvm-ipa-wholedisk-bios-agent_ipmitool-tinyipa&i=nope&files=&repos=project-config15:45
openstackgerritIlya Etingof proposed openstack/sushy-tools master: Add memoization to expensive emulator calls  https://review.openstack.org/61275815:46
NobodyCamGood Morning and happy hump day15:50
iurygregorygood morning NobodyCam o/15:51
dtantsurmorning NobodyCam15:51
NobodyCamO/ iurygregory and dtantsur :)15:51
rpittauhi NobodyCam :)15:54
NobodyCamGood Morning rpittau :)15:55
TheJuliasomething is super fishy about this failure :( and no sushy involved15:56
TheJulialikely, that is performance graphing someone setup at some point...15:57
TheJuliadoh, its a qcow, not raw image15:59
TheJuliaokay, after converting16:00
TheJuliayes, the 0.4.0 is a gpt filesystem16:00
dtantsur\o/16:03
dtantsurso we've lost coverage for BIOS+MBR? :D16:03
*** pcaruana has quit IRC16:03
TheJuliafor wholedisk, yes16:03
TheJuliabut gpt is fine16:03
TheJuliaerr partition based is fine16:04
TheJuliawe have to trust the image" and we are doing it in whole disk jobs16:04
TheJuliawe're just much more resilliant apparently16:04
dtantsuryeah, but GPT and MBR are different code paths even for wholedisk images (I think)16:04
TheJuliadtantsur: interleaved16:04
dtantsuryeah. in quite some spaghetti :)16:05
TheJuliaindeed16:06
TheJuliaI'm actually really happy that it passed16:06
TheJuliawell16:06
TheJuliagets through16:06
TheJuliait fails for other reasons it seems16:06
* TheJulia gets out the colander16:09
* TheJulia wonders if anyone uses the partial code for local caching for nodes awaiting official deployment orders16:13
dtantsurgood question16:14
*** gyee has joined #openstack-ironic16:18
*** moshele has joined #openstack-ironic16:24
TheJuliasomething is really not adding up for me, maybe I just need more coffee16:27
dtantsursomething? I'm completely in darkness with this issue :) if "just throw another partprobe" does not work, I dunno..16:28
TheJuliaI think that is the only option, starting to understand16:30
*** moshele has quit IRC16:32
* TheJulia opens up parted's source code to confirm theory16:32
dtantsurwow16:32
*** rnoriega has quit IRC16:38
*** radez has quit IRC16:38
openstackgerritMerged openstack/sushy master: Add System<->Manager linkage follow up  https://review.openstack.org/62950016:38
TheJuliasweet... github failed16:39
TheJuliaoh, its back now :(16:39
*** ltomasbo has quit IRC16:39
*** lucasagomes has quit IRC16:39
TheJuliawow, the way it is written makes my head hurt16:40
iurygregorydtantsur, should i trigger the check-experimental in nova since some ironic jobs run only there? (2 jobs are in experimental) https://review.openstack.org/#/c/629504/1/.zuul.yaml16:41
patchbotpatch 629504 - nova - Rename Ironic jobs - 1 patch set16:41
*** lucasagomes has joined #openstack-ironic16:41
*** rnoriega has joined #openstack-ironic16:43
TheJuliayeah, I suspect my feeling is right16:44
*** lucasagomes has quit IRC16:45
*** lucasagomes has joined #openstack-ironic16:50
dtantsuriurygregory: yeah, you can try it16:51
dtantsurTheJulia: so?16:51
iurygregorydtantsur, ack tks o/16:51
TheJuliastill sifting through the c code16:51
TheJuliaI just confirmed it should be fsyncing, so it is likely just that, the kernel is not loading the table afterwards.   There is a note on the fsync docs about behavior dependant upon filesystem though...16:53
dtantsur"udevadm settle" returned: 0 in 38.853s  <- impressive16:54
TheJuliawow16:54
* TheJulia saves other words for not online16:54
dtantsurI see. I'm looking at http://logs.openstack.org/47/629547/1/check/ironic-lib-wholedisk-bios-ipmi-direct-tinyipa-src-python2/f7961c8/controller/logs/ironic-bm-logs/node-0_no_ansi_2019-01-09-16:28:34_log.txt.gz#_2019-01-09_16_28_03_502 I cannot see my partprobe. weird.16:55
TheJuliabecause we don't build it in by default afaik16:55
TheJuliaoh, did we build ipa from scratch with the code?16:55
dtantsurTheJulia: well, we used to build IPA in ironic-lib gate at least16:56
dtantsurbut I have a feeling it got broken. sigh. SIGH.16:57
TheJuliaugh17:01
dtantsuroh, I see. we hardcode /opt/stack/new in some places. it's no longer the thing..17:03
*** e0ne has quit IRC17:03
* TheJulia gets out whiskey17:03
TheJuliafsync on a ramdisk is an explicit noop17:03
TheJuliait does nothing17:03
*** robbbe has quit IRC17:03
TheJuliahttps://github.com/torvalds/linux/blob/6f0d349d922ba44e4348a17a78ea51b7135965b1/fs/ramfs/file-mmu.c17:04
rpittauI was going to ask a couple of things but I see the whiskey level is already high enough, it can wait :P17:04
NobodyCamrpittau: LOL17:04
TheJuliaso io data from parted is in the memory cache, and waits upon the system to do lazy writing out since it doesn't appear to do explicit direct io calls17:05
TheJuliareading however for the kernel requires a lock from the device, so there is the case without forcing a disk sync that the new partition is in ram, but not yet on disk when we attempt to rescan17:06
rpittaubye all! good evening o/17:08
NobodyCamhave a good night rpittau17:08
*** rpittau has quit IRC17:08
openstackgerritDmitry Tantsur proposed openstack/ironic-python-agent master: Recover CI run detection in imagebuild  https://review.openstack.org/62957417:09
dtantsurTheJulia: maybe this can help ^^^ but I'm too tired to think more17:09
TheJuliadtantsur: I think we need to perform an explicit sync as well to totally eliminate slow disk io being an isue17:09
dtantsurTheJulia: in my ironic-lib patch?17:10
TheJuliathat will force the kernel to go disk by disk syncing out buffers17:10
TheJuliadtantsur: yeah, I can do it17:10
dtantsuryeah, feel free to update it17:10
*** dustinc has joined #openstack-ironic17:10
TheJuliak17:10
TheJuliacoffee first17:10
* etingof has been wondering about explicit sync17:10
dtantsursee you17:10
*** dtantsur is now known as dtantsur|afk17:11
TheJuliayeah, seems silly to ahve to do it, but ramdisk the rules are different17:11
TheJuliagoodnight17:11
* etingof observed that sysadmins tend to run `sync` before committing into some serious business17:12
etingoffelt like a genetic trait17:12
TheJuliayeah, provides some guarantee (minus raid controllers lying about io cache flush-out) that things are written to disk17:16
TheJuliain case, oh... accidental power button or the like17:16
TheJuliaI had a database platform I was responsible for the datacenter operations and system administration of a long time ago, and it used a mix of direct io and let the rest of the io be a lazy write to disk. The aspects that were a lazy write are almost always the part that led to corruption17:18
TheJuliaand it was possible to also crash some of the raid controllers then with massive direct io queues17:19
TheJuliasoooo many database restores....17:19
etingofso the habit of occasional syncing is a good trait ;)17:19
*** moshele has joined #openstack-ironic17:19
TheJuliaindeed17:23
openstackgerritMerged openstack/networking-generic-switch master: Directly source ML2 plugin for devstack  https://review.openstack.org/62870817:28
*** moshele has quit IRC17:28
TheJuliaInteresting dmitry's wip patch failed pep8 :\ but is fine locally17:31
TheJuliaoh, no, it did17:31
TheJuliawrong line \o/17:31
openstackgerritDerek Higgins proposed openstack/ironic master: [WIP] Replace use of Q_USE_PROVIDERNET_FOR_PUBLIC  https://review.openstack.org/62114617:49
*** dougsz has quit IRC17:54
*** trown is now known as trown|lunch17:55
*** derekh has quit IRC17:56
*** MattMan has quit IRC18:11
*** MattMan has joined #openstack-ironic18:11
*** dustinc has quit IRC18:22
*** dustinc has joined #openstack-ironic18:23
*** S4ren has quit IRC18:30
*** e0ne has joined #openstack-ironic18:35
openstackgerritJulia Kreger proposed openstack/ironic-lib master: Run sync and partprobe after adding a configdrive partition  https://review.openstack.org/62954718:44
*** e0ne has quit IRC19:02
openstackgerritHamdy Khader proposed openstack/ironic master: Add is_smartnic to Port data model  https://review.openstack.org/62964019:04
*** e0ne has joined #openstack-ironic19:05
*** trown|lunch is now known as trown19:08
*** moshele has joined #openstack-ironic19:50
*** e0ne has quit IRC20:08
*** dustinc has quit IRC20:37
*** S4ren has joined #openstack-ironic20:51
*** dustinc has joined #openstack-ironic20:52
openstackgerritJulia Kreger proposed openstack/ironic master: WIP ipv6 job  https://review.openstack.org/61316921:12
TheJuliaCore reviews on https://review.openstack.org/#/c/629574/1 would be appreciated21:16
patchbotpatch 629574 - ironic-python-agent - Recover CI run detection in imagebuild - 1 patch set21:16
*** moshele has quit IRC21:29
openstackgerritJulia Kreger proposed openstack/ironic master: WIP ipv6 job  https://review.openstack.org/61316921:57
*** baha has quit IRC22:01
*** trown is now known as trown|outtypewww22:10
TheJuliaand core reviews on https://review.openstack.org/#/c/629547/ would also be appreciated22:31
patchbotpatch 629547 - ironic-lib - Run sync and partprobe after adding a configdrive ... - 2 patch sets22:31
*** rcernin has joined #openstack-ironic22:52
*** jcoufal has quit IRC23:03
jidarcan I re-run the undercloud installer (openstack undercloud install) and move the interface over from say, eno3 to eno2 if I messed it up?23:22
*** dustinc has quit IRC23:33
*** dustinc has joined #openstack-ironic23:36
« Tuesday, 2019-01-08 Index Thursday, 2019-01-10 »

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!