Wednesday, 2017-12-06

« Tuesday, 2017-12-05 Index Thursday, 2017-12-07 »
*** btully has joined #openstack-glance00:15
*** btully has quit IRC00:20
*** threestrands has quit IRC00:25
*** gyee has quit IRC01:10
*** markvoelker has quit IRC01:38
*** openstackgerrit has joined #openstack-glance01:44
openstackgerritDamon Li proposed openstack/glance master: Add the list of hw_version supported by vmware driver.  https://review.openstack.org/52584501:44
*** zhurong has joined #openstack-glance01:49
*** namnh has joined #openstack-glance01:53
*** dalgaaf has quit IRC02:01
*** dalgaaf has joined #openstack-glance02:02
*** btully has joined #openstack-glance02:04
*** btully has quit IRC02:09
*** pbourke_ has quit IRC02:25
*** pbourke_ has joined #openstack-glance02:27
*** markvoelker has joined #openstack-glance02:39
*** zhurong has quit IRC03:02
*** MattMan has quit IRC03:10
*** zhurong has joined #openstack-glance03:10
*** MattMan has joined #openstack-glance03:10
openstackgerritChason Chan proposed openstack/glance master: Update the valid disk bus list for qemu and kvm hypervisors  https://review.openstack.org/51389403:16
*** abhishekk has joined #openstack-glance03:29
openstackgerritAbhishek Kekane proposed openstack/glance master: Fix 500 on ValueError during image-import  https://review.openstack.org/52270803:50
openstackgerritBrian Rosmaita proposed openstack/glance master: [import-tests] adds tests for image-import/staging  https://review.openstack.org/52302803:53
bhagyashrisrosmaita: Hi, I have submitted follow-on patch for add 'stage_image' policy could you please take some insights https://review.openstack.org/#/c/525578/ also kairat has given comment on https://review.openstack.org/#/c/524060/ so what is your opinion about it03:55
rosmaitayeah, i saw kairat's comment, he's technically correct, need to think about that some more03:56
bhagyashrisrosmaita: actually set_data() method is common for both the import and upload case and also policy name is hard coded03:57
bhagyashrisrosmaita: and that is used in lot many places as well03:57
bhagyashrisrosmaita: yeah03:58
*** links has joined #openstack-glance03:59
*** links has quit IRC04:01
*** links has joined #openstack-glance04:01
*** Dinesh_Bhor has joined #openstack-glance04:10
*** zhurong has quit IRC04:21
*** trungnv has quit IRC04:41
*** namnh has quit IRC04:41
*** trungnv has joined #openstack-glance04:42
*** namnh has joined #openstack-glance04:42
*** udesale has joined #openstack-glance04:46
*** sapd_ has quit IRC04:48
*** sapd has joined #openstack-glance04:50
rosmaitabhagyashris i put a not very helpful comment on https://review.openstack.org/#/c/524060/04:52
rosmaitawe need to see what erno, abhishek, and nikhil think04:52
openstackgerritBrian Rosmaita proposed openstack/glance master: Fix 500 from duplicate stage call  https://review.openstack.org/52145704:53
bhagyashrisrosmaita: thank you for your for comment i would also like to see other opinion :)04:54
*** ratailor has joined #openstack-glance04:58
*** Dinesh_Bhor has quit IRC05:01
*** ratailor has quit IRC05:05
*** ratailor has joined #openstack-glance05:08
*** zhurong has joined #openstack-glance05:09
*** threestrands has joined #openstack-glance05:09
*** threestrands has quit IRC05:09
*** rosmaita has quit IRC05:26
*** tshefi has joined #openstack-glance05:47
*** Dinesh_Bhor has joined #openstack-glance06:00
*** AlexeyAbashkin has joined #openstack-glance06:07
*** udesale__ has joined #openstack-glance06:10
*** AlexeyAbashkin has quit IRC06:11
*** AlexeyAbashkin has joined #openstack-glance06:12
*** udesale has quit IRC06:12
*** AlexeyAbashkin has quit IRC06:26
*** pdeore has joined #openstack-glance06:26
openstackgerritGuoqiang Ding proposed openstack/glance master: Update the documentation links  https://review.openstack.org/52328606:33
*** bkopilov has quit IRC06:35
*** arcolife has joined #openstack-glance06:37
*** tshefi has quit IRC06:40
*** tshefi has joined #openstack-glance06:41
*** mosulica has joined #openstack-glance06:55
*** udesale has joined #openstack-glance06:58
*** udesale__ has quit IRC07:00
*** rcernin has quit IRC07:02
*** e0ne has joined #openstack-glance07:06
*** tshefi has quit IRC07:29
*** btully has joined #openstack-glance07:30
*** arcolife has quit IRC07:33
*** btully has quit IRC07:35
*** rcernin has joined #openstack-glance07:39
*** arcolife has joined #openstack-glance08:01
*** lin_yang has quit IRC08:10
*** alexchadin has joined #openstack-glance08:14
*** tesseract has joined #openstack-glance08:19
*** e0ne has quit IRC08:32
*** AlexeyAbashkin has joined #openstack-glance08:32
*** zhurong has quit IRC08:33
*** AlexeyAbashkin has quit IRC08:38
*** udesale has quit IRC08:42
*** AlexeyAbashkin has joined #openstack-glance08:44
*** AlexeyAbashkin has quit IRC08:49
*** zhurong has joined #openstack-glance08:53
*** Dinesh_Bhor has quit IRC09:09
*** AlexeyAbashkin has joined #openstack-glance09:14
*** d0ugal has quit IRC09:37
*** d0ugal has joined #openstack-glance09:41
*** e0ne has joined #openstack-glance09:48
openstackgerritliuyamin proposed openstack/glance master: doc migrate: update the doc link address  https://review.openstack.org/50778409:54
*** linkmark has joined #openstack-glance09:57
*** lin_yang has joined #openstack-glance10:08
*** namnh has quit IRC10:25
*** zhurong has quit IRC10:37
*** d0ugal has quit IRC10:39
*** alexchadin has quit IRC10:48
*** mvk has quit IRC10:52
*** udesale has joined #openstack-glance10:55
*** tshefi has joined #openstack-glance10:59
*** abhishekk has quit IRC11:00
*** d0ugal has joined #openstack-glance11:01
*** udesale has quit IRC11:20
*** links has quit IRC11:34
*** rcernin has quit IRC11:39
*** rosmaita has joined #openstack-glance11:40
*** ratailor has quit IRC11:47
*** links has joined #openstack-glance11:48
*** pdeore has quit IRC11:48
*** alexchadin has joined #openstack-glance11:49
*** AlexeyAbashkin has quit IRC11:58
*** pdeore has joined #openstack-glance12:01
*** openstackgerrit has quit IRC12:03
*** links has quit IRC12:08
*** links has joined #openstack-glance12:13
*** alexchadin has quit IRC12:16
*** alexchadin has joined #openstack-glance12:16
*** btully has joined #openstack-glance12:21
*** btully has quit IRC12:32
*** mvk has joined #openstack-glance12:41
*** AlexeyAbashkin has joined #openstack-glance12:46
*** tesseract has quit IRC12:53
*** tesseract has joined #openstack-glance12:57
*** takedakn has joined #openstack-glance13:09
rosmaitajokke_ : need to talk to you for a few min in your capacity as release czar13:17
*** links has quit IRC13:23
*** markvoelker has quit IRC13:24
*** markvoelker has joined #openstack-glance13:25
*** openstackgerrit has joined #openstack-glance13:28
openstackgerritMonty Taylor proposed openstack/python-glanceclient master: Update legacy jobs to use shiny new devstack job  https://review.openstack.org/52606413:28
openstackgerritMonty Taylor proposed openstack/python-glanceclient master: Update legacy jobs to use shiny new devstack job  https://review.openstack.org/52606413:34
*** takedakn has quit IRC13:39
*** takedakn has joined #openstack-glance13:41
*** alexchadin has quit IRC13:45
*** arcolife has quit IRC14:05
*** kuzko has quit IRC14:07
*** kuzko has joined #openstack-glance14:09
*** udesale has joined #openstack-glance14:12
*** pbourke_ is now known as pbourke14:12
*** takedakn has quit IRC14:18
rosmaitajokke_ i have updated the q-2 release etherpad14:32
rosmaitahttps://etherpad.openstack.org/p/glance-queens-Q214:32
rosmaitaquick question before we go through the items14:33
rosmaitado you have any objections to the IIR test coverage additions?14:33
rosmaitahttps://review.openstack.org/#/c/523028/14:33
rosmaitahttps://review.openstack.org/#/c/523179/14:33
jokke_not really, have been happy to see them passing on check. 'though I don't think it's the best use for gating time for the tomorrows release14:34
rosmaitai am tempted to ninja them in on the theory that more test coverage is always good ... the problem with that theory is, "more" might be taken to mean "sufficient", and that might not necessarily be the case14:34
rosmaitawell, that's a good point, though i am tired of rebasing/merge resolving them!14:35
jokke_so I've been focusing on the bugfixes including tests etc, but the new tests does not really affect the release itself14:35
jokke_that is fair enough ;)14:35
rosmaitaok, makes sense14:35
rosmaitalet's take a look at the scrubber patch, specifically https://review.openstack.org/#/c/510449/11/glance/scrubber.py@21414:36
jokke_so how about we push them in first thing tomorrow after the rush has calmed for Q2?14:36
rosmaitaworks for me, i will put a note14:36
*** tesseract has quit IRC14:37
jokke_yeah, I can +A them tomorrow morning when I'm back to work :D14:38
rosmaitaor friday if we want the gate clear for other people ... seems to be running slow14:39
jokke_yeah lets see ... tomorrow evening at least should be calming down already as the Q2 should be tagged tomorrow so I expect most projects being out of the rush mode14:40
rosmaitabtw, monty is helping out on the glanceclient functional test situation, he's got a patch up14:41
jokke_nnice14:42
jokke_so the scrubber14:42
rosmaitaright14:42
jokke_I don't see any reason to use that config option (it's definitely misleading) nor in matter of fact do any paging on that db query what so ever14:43
*** tesseract has joined #openstack-glance14:44
rosmaitawell, i don't know that we need to grab everything at once14:44
jokke_we're pulling only pending delete rows from the db ... even in the odd case that someone has been really really active and we get 1000 line return we actually put way more load on the db by quering it 40 time (like te value there with 25 default) than just getting the results14:45
rosmaitabut the limit could prob be 500 instead of 2514:45
jokke_it's not like we're going through any http api there anymore with heavy parsing14:45
rosmaitatrue14:45
rosmaitalet's say hard code 750 in there14:46
jokke_like the amount of data that's coming from db at that point is quite minimal and I'd say any paging/limiter there is rather pushing the performance down than beneficial14:46
jokke_id say lets not put anything there and if someone finds a case where such would be actually needed we can revisit that14:47
rosmaitawell, key thing is the patch is unacceptable using that config option ATM14:47
rosmaita(i will check with a DB expert about whether we need to page or not in the meantime)14:47
jokke_yes so lets remove that paging all together14:47
jokke_if someone has 50k pending delete images waiting for scrubber, I think pulling 50k images from db is least of their issues :P14:48
jokke_and it's not like scrubber is doing anything heavy with that data. I just can't see a case where paging that db query would be beneficial14:49
rosmaitayeah, but i think there are some people running delayed delete on a long window so they can undelete images14:49
jokke_tru and that still applies14:49
jokke_it's not like scrubber is going to issue delete on all of them to the backend at the same time. it just builds the list at that point14:50
jokke_the throttling is taken care of way later in the process14:50
rosmaitaok, how about https://review.openstack.org/#/c/522708/14:50
rosmaitai guess i'm worried about OOM error building a huge list14:51
rosmaitai think this wasn't the one you had a reservation about yesterday?14:52
jokke_so I've been trying to understand those 2 patches but we just have been running across eachother with Abhishek14:53
rosmaitayeah, i saw your comment on the other one about how could this one happen14:54
jokke_so there is this one and one that prevents the image going active if those values are missing14:54
jokke_yeah14:54
jokke_and I haven't wanted to really give green light ot either one of these before understanding how we manage to get to the point where we barf on the import call returning 500 and yet still apparently activate the image14:55
jokke_I'd like to understand is fixing one of these causing the other14:55
jokke_because in that case I think we need to squash them into one PS and make sure we do not allow that to happen14:56
rosmaitaok, so the "other one" is https://review.openstack.org/#/c/522774/14:58
jokke_so next one ... 524060 ... while I agree with Kairat that it does contradict with the policy layer architecture, I also think that's one of the dumbest things in there and would be happy to move all those policies out of there and delete that layer all together because that architecture contradicts with all the policy enforcement principles in OpenStack like we have discussed multiple times before14:59
jokke_ ;)14:59
rosmaitaright, i also think it's the thing blocking image location reform15:00
jokke_I tihnk so too15:00
rosmaitai don't think we want to introduce a controller proxy and put the enforcement in the policy layer15:01
jokke_so yes that 522774 is the counterpart for the previous one15:01
jokke_let me put comment into that15:01
jokke_rosmaita: I will happily -2 any such proposed patches15:01
rosmaitayeah, i'm pretty sure i saw 522774 when testing at the end of pike, i believe that you can't update the container_format or disk_format on an active image15:02
rosmaitaso that was definitely a problem15:02
rosmaitathat would be good to have fixed15:02
jokke_so I don't understand what the problem is there on that 52277415:02
jokke_like Abhishek says it doesn't come active without the dontainer and disk formats which is the exact case that patch is addressing so it's not changing anything15:03
jokke_it already does not behave like the bug claims15:03
rosmaitawait15:04
rosmaitaok, two bugs15:04
rosmaita(1) if no c_f, d_f set, the image is in 'uploading' status, you can call import, the image goes active without c_f, d_f15:05
rosmaita(2) no c_f, d_f set, image in 'queued' status, you call import, get 50015:05
jokke_ohhhh ... that's the part I missed15:05
jokke_so yes ... then these 2 different fixes are needed15:06
rosmaitaok, cool15:06
rosmaitai think the info is clearer in the bug reports15:07
jokke_although the fix for queued will need to change again onvce we get the copy-from replacement into place and we need to reject the call based on the body of the import call15:07
jokke_now it's just dumbly rejecting it, which might be fine for now, but makes life more difficult in the future15:07
rosmaitaright15:08
jokke_I'm almost tempted to say let it return 500 for now and lets do it properly when we need the logic into the place to handle those multiple situation which needs to happen by Q3 anyways15:09
jokke_happy to get the active without c_f & d_f in for Q2 'though now when I understand the mechanims for them15:09
jokke_I'll +A that as you have your +2 there already15:11
rosmaitai see your point about the 500, but on the other hand, i hate 500s15:11
*** belmoreira has joined #openstack-glance15:11
jokke_me too and if this was on stable API I wouldn't hesitate a second to get it fixed on milestone release15:12
rosmaitaon https://review.openstack.org/#/c/522774/, do you see that little orange circle next to "Parent(s)" ?15:12
rosmaitai think andreas told me the other day that's an indicator that the patch needs rebasing15:13
jokke_oh15:13
rosmaitathough i may have misunderstood him15:13
rosmaitajust wondering if you had noticed anything like that?15:13
jokke_oh F*ck15:14
jokke_the 522774 is depending on 522708 ... I just hate gerrit as it is now15:15
rosmaitacrap15:15
rosmaitawell, let's go with "yes" on 52270815:15
jokke_oh it's different change anyways15:17
rosmaitamaybe that's what adreas said, that an orange circle means the parent is unmerged15:17
jokke_that's fine to approve15:17
rosmaitaok, cool15:18
jokke_ok so 52302915:20
jokke_I have not been able to dig into that yet. I think for the patch to work the file store needs to be enabled but I'm not 100% sure15:21
jokke_our store initializations are kind of mess15:21
jokke_if that's the case I don't think we should merge that as is15:21
rosmaitai was wondering about that, it is very file-store-specific15:22
jokke_yes and to initialize the staging area we had to do quite nasty hacks to make it possible and that is directly on the staging code alone15:22
jokke_so I'm bit hesitant to approve that as I'm not sure about the side effects15:23
jokke_I'm also not sure how that behaves if you do it during the staging upload still being running15:24
rosmaitagood point15:24
rosmaitai will move that bug to Q-3 and put the above comments in it15:24
jokke_thnx15:24
jokke_anything else for today?15:25
rosmaitai guess that's it ... what should we do about the policy patch?15:27
jokke_unless you wanna fix it like in next few mins I quess it will be pushed back to Q3 :P15:28
rosmaitaright15:28
jokke_I'm planning to continue 100th independence day preparations :D15:28
jokke_which is today15:28
rosmaitawas just going to ask that15:29
rosmaitahappy independence day15:29
rosmaitai'm going to ask around about the DB paging thing, may revise that patch myself15:29
rosmaitai'd like to get the new scrubber into q-215:30
jokke_sure15:34
rosmaitaok, will leave you a note if i do15:35
jokke_ping me on telegram if you need me still today15:35
jokke_I have no problem hopping on if needed, but not gonna follow up irc15:36
jokke_nor mail15:36
*** udesale has quit IRC15:36
rosmaitaif i make the change, and abhishek likes it, you can approve when you get in in the morning15:37
rosmaita(if you approve)15:37
*** mvk has quit IRC15:48
openstackgerritMerged openstack/glance master: Utilize LimitingReader for staging data  https://review.openstack.org/52141115:59
*** btully has joined #openstack-glance16:06
*** btully has quit IRC16:10
*** belmoreira has quit IRC16:22
*** mosulica has quit IRC16:23
*** gyee has joined #openstack-glance16:48
*** tshefi has quit IRC16:51
*** pdeore has quit IRC16:53
*** AlexeyAbashkin has quit IRC16:54
*** e0ne has quit IRC17:07
*** abhishekk has joined #openstack-glance17:54
*** abhishekk has quit IRC17:54
*** btully has joined #openstack-glance17:55
*** abhishekk has joined #openstack-glance17:58
*** btully has quit IRC18:00
*** tesseract has quit IRC18:07
openstackgerritAbhishek Kekane proposed openstack/glance master: Fix 500 from duplicate stage call  https://review.openstack.org/52145718:12
*** abhishekk has quit IRC18:22
*** e0ne has joined #openstack-glance18:39
*** harlowja has quit IRC19:32
*** harlowja has joined #openstack-glance19:36
*** mvk has joined #openstack-glance19:39
*** Anticime1 has quit IRC20:31
*** jdurgin has quit IRC20:31
*** gyee has quit IRC20:31
*** rosmaita has quit IRC20:31
*** Anticime1 has joined #openstack-glance20:47
*** jdurgin has joined #openstack-glance20:47
*** gyee has joined #openstack-glance20:47
*** rosmaita has joined #openstack-glance20:47
*** harlowja has quit IRC21:00
*** harlowja has joined #openstack-glance21:03
*** harlowja has quit IRC21:32
*** AlexeyAbashkin has joined #openstack-glance21:38
*** openstackgerrit has quit IRC21:38
*** lifeless has quit IRC21:38
*** mfisch has quit IRC21:38
*** diegows has quit IRC21:38
*** jinli has quit IRC21:38
*** openstackgerrit has joined #openstack-glance22:01
*** lifeless has joined #openstack-glance22:01
*** mfisch has joined #openstack-glance22:01
*** diegows has joined #openstack-glance22:01
*** jinli has joined #openstack-glance22:01
*** rcernin has joined #openstack-glance22:01
*** rcernin has quit IRC22:03
*** rcernin has joined #openstack-glance22:03
*** e0ne has quit IRC22:06
*** e0ne has joined #openstack-glance22:07
*** e0ne has quit IRC22:07
*** e0ne has joined #openstack-glance22:08
*** e0ne has quit IRC22:08
*** e0ne has joined #openstack-glance22:08
*** e0ne has quit IRC22:09
*** e0ne has joined #openstack-glance22:10
*** e0ne has quit IRC22:10
*** e0ne has joined #openstack-glance22:11
*** e0ne has quit IRC22:11
*** e0ne has joined #openstack-glance22:12
*** e0ne has quit IRC22:12
*** e0ne has joined #openstack-glance22:12
*** e0ne has quit IRC22:13
*** harlowja has joined #openstack-glance22:36
*** AlexeyAbashkin has quit IRC22:43
*** openstack has joined #openstack-glance23:27
*** ChanServ sets mode: +o openstack23:27
*** takedakn has joined #openstack-glance23:30
*** takedakn has quit IRC23:34
« Tuesday, 2017-12-05 Index Thursday, 2017-12-07 »

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