15:00:32 <bswartz> #startmeeting manila 15:00:33 <openstack> Meeting started Thu Jan 25 15:00:32 2018 UTC and is due to finish in 60 minutes. The chair is bswartz. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:34 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:37 <openstack> The meeting name has been set to 'manila' 15:00:42 <amito> o/ hello 15:00:45 <bswartz> hello all 15:00:48 <dustins> \o 15:00:55 <gouthamr> hello o/ 15:00:57 <ganso> hello 15:01:06 <zhongjun> Hi 15:01:06 <xyang> hi 15:01:23 <tbarron> hi 15:02:37 <bswartz> hey guys, sorry I'm dropping a -2 on some patches 15:03:25 <bswartz> Okay! 15:03:26 <tbarron> thud 15:03:34 <bswartz> #topic announcements 15:03:47 <bswartz> Today is the last day for features to merge 15:03:52 <jungleboyj> @! 15:03:53 <bswartz> I need to push tags today 15:04:12 <bswartz> jungleboyj: your bot is awol 15:04:22 * jungleboyj is sad 15:04:53 <bswartz> I don't think we have an other announcements 15:05:11 <bswartz> Feature freeze is usually the biggest deadline 15:05:24 <bswartz> #agenda https://wiki.openstack.org/wiki/Manila/Meetings 15:05:31 <bswartz> #topic Feature Freeze / Queens milestone-3 15:05:46 <bswartz> So we've been trying to get the last feature patches merged this week 15:06:01 <bswartz> There's been good progress, but this morning I still see some unmerged patches, so let's go through them 15:06:47 <bswartz> The filtering for the share type API won't make it, so I've already block those 2 patches 15:07:29 <bswartz> We have this one waiting for a workflow: https://review.openstack.org/#/c/527134/ 15:08:02 * tbarron is timing out connecting to gerrit atm 15:08:09 <gouthamr> was testing that one, need xyang/tbarron to weigh in 15:08:16 <gouthamr> or zhongjun 15:08:35 <bswartz> gouthamr: found any issues worth discussing? 15:08:47 <gouthamr> nope 15:09:30 <bswartz> https://review.openstack.org/#/c/528667/ 15:09:52 <bswartz> ^ This one looks like it's suffering infra problems 15:09:53 <gouthamr> ouch, rechecking 15:10:19 <bswartz> https://review.openstack.org/#/c/535701/ 15:10:29 <bswartz> ^ This is a 5 line change 15:10:36 <efried> No rechecking 15:10:44 <bswartz> Both of its dependencies are merged 15:10:48 <efried> Wait for next infra broadcast 15:10:52 <bswartz> efried: ? 15:11:13 <tbarron> bswartz: 535701 is ok, I was just waiting for the dependencies 15:11:14 <efried> NOTICE: We're currently experiencing issues with the logs.openstack.org server which will result in POST_FAILURE for jobs, please stand by and don't needlessly recheck jobs while we troubleshoot the problem. 15:11:15 <bswartz> efried: is the recheck machinery broken? 15:11:47 * bswartz smh 15:11:52 <gouthamr> https://review.openstack.org/#/c/535701/ is failing a bunch of third party CIs 15:12:32 <tbarron> thirdy party CIs are failing for another reason 15:12:36 <tbarron> third 15:12:37 <bswartz> ganso: can you speak to that? 15:12:44 <tbarron> including NetApp 15:13:13 <ganso> bswartz: erlon and I are diagnosing a CI failure due to library update 15:13:14 <gouthamr> looks like a broken package atleast for NTAP, python-pcre 15:13:19 <tbarron> python-pcre paacke issue 15:13:25 <ganso> bswartz: erlon says all CIs are failing 15:13:25 <tbarron> http://52.42.67.99/67/537867/1/check/dsvm-veritas-access-manila-driver/95ec234/logs/devstack-early.txt.gz#_2018-01-25_12_22_34_363 15:13:36 <tbarron> ganso: all 3rd party 15:13:39 * bswartz headdesk 15:13:41 <tbarron> ganso: not infra 15:14:12 <tbarron> infra jobs are working, ubuntu and centos, but ubuntu 3rd party has the issue 15:14:17 <erlon_> tbarron, yep infra not, https://review.openstack.org/#/c/534854/ 15:14:28 <erlon_> but 90% of the tird party 15:14:31 <bswartz> Okay, this is a perfect illustration of why it's better to get stuff merged on Monday or Tuesday, not Thursday 15:14:35 <erlon_> not sure why 15:14:44 <bswartz> Library problems seem to happen every release 15:14:58 <bswartz> Is it something we can get sorted out this morning? 15:15:24 <bswartz> Or will we have to accept that third party CI is broken until after FF? 15:15:26 <tbarron> https://review.openstack.org/#/c/537867/ is a DNM job that illustrates 15:15:46 <erlon_> bswartz, not sure, couldnt get help from #infra as they are focusing on the logserver problem 15:15:59 <tbarron> infra jobs are all passing, third party are all failing 15:16:18 <ganso> tbarron: not really passing, they have POST_FAILURE issue now 15:16:22 <bswartz> #link https://pypi.python.org/pypi/python-pcre/0.6 15:16:23 <bswartz> this? 15:16:31 <gouthamr> infra has mirrors, they'll be affected late :) 15:16:41 <erlon_> ganso, they are going further at least :p 15:17:22 <tbarron> ganso: they passed in 537867 two hours ago, iow they don't have the package issue 15:17:42 <tbarron> gouthamr: ah, maybe just delayed 15:18:04 <bswartz> I don't understand what's changed 15:18:15 <bswartz> That package hasn't been modified since 2015 according to pypi 15:19:11 <erlon_> bswartz, Im not getting the problem as well, the bindep is already there, for long time: https://github.com/openstack/requirements/search?utf8=%E2%9C%93&q=pcre&type= 15:19:47 <erlon_> so, devstack or something else should know that it should be installed before trying to install the python lib 15:20:12 <bswartz> I'm confused 15:20:16 <tbarron> did we always have to compile it? 15:20:50 <bswartz> tbarron -- if you install from pip, I think the native parts get compiled as part of the install 15:20:51 <tbarron> looks like the dev pkg might be missing, the compile fails for lack of pcre.h 15:21:12 <erlon_> tbarron, not sure, may be that was what is different now, now it needs to be compiled 15:21:14 <bswartz> If you install from apt/yum you get a distro-compile binary 15:21:23 <tbarron> bswartz: ack, understood 15:22:00 <bswartz> It's possible the pcre-dev package moved some files and the python package is now broken 15:22:15 <bswartz> So there may be a python-pcre 0.7.1 very soon 15:22:19 <erlon_> bswartz, that will download the pcre lib only, the pip package still need to compile 15:22:58 <bswartz> I'm okay with ignoring the 3rd party CIs until that issue is sorted 15:23:23 <bswartz> It sounds like the upstream gate might be about to get less friendly for the same reasons, so we should hurry up and merge what we can 15:24:16 <bswartz> Okay are there any other patches I missed? 15:24:24 <bswartz> For Q-3? 15:24:45 <bswartz> If not, I'm going to watch for the last few things to get through the gate and push tags as soon as they make it 15:25:18 <tbarron> looking at devstack-early log in a successful third party CI run from a couple weeks ago I see no mention of python-pcre 15:27:17 <bswartz> I know the netapp guys will dig into to try to fix the netapp CI at least 15:27:27 <bswartz> Please share any fixes you find on the channel 15:27:45 <bswartz> And lmk if you want my help on that CI issue 15:28:05 <bswartz> Okay I think that's it for feature freeze 15:28:18 <tpsilva> a quick workaround we found is to install libpcre3-dev on the nodes before stacking 15:28:31 <bswartz> tpsilva: ty 15:29:01 <bswartz> #topic Let's Go Over New Bugs 15:29:40 <bswartz> Before we get into specific bugs, let me remind y'all that the focus between feature freeze and RC1 should be on finding and fixing bugs 15:30:00 <bswartz> There are some bugfix patches already waiting for reviews 15:30:23 <bswartz> I will be targeting bugs at the RC1 milestone in LP 15:30:57 <bswartz> Arguably, finding bug is as important as fixing the known bugs, so I encourage people to test out some of the new features and try to break them 15:31:52 <bswartz> The RC1 target date for us is 2 weeks from now 15:32:26 <bswartz> As always, we'll cut the RC1 tag when the number of bugs drops to zero, whether that's early or late 15:32:53 <bswartz> But the closer we get to the target date, the more aggressively we'll untarget bugs -- our release manager don't like late RCs 15:33:00 <bswartz> :-) 15:33:08 <bswartz> dustins: do you have specific bugs for today? 15:33:16 <dustins> I do indeed 15:33:19 <dustins> #link https://bugs.launchpad.net/manila/+bug/1550258 15:33:20 <openstack> Launchpad bug 1550258 in Manila "Manila UI - Manage rule for CIFS share updates status from "error" to "active" for invalid users" [Low,New] 15:33:27 <bswartz> #link https://etherpad.openstack.org/p/manila-bug-triage-pad 15:34:04 <bswartz> Wow this is an old one 15:34:41 <bswartz> Why did I target this to manila-ui? 15:34:55 <gouthamr> it's marked invalid on manila-ui 15:34:59 <bswartz> It must have been a click error 15:35:19 <bswartz> In any case, is it HP3par specific? 15:35:33 <tbarron> there was a ui screenshot in the bug report but then it was realized that it's not a ui bug 15:35:42 <dustins> It looks like it manifested first in the UI but is caused by a Manila "core" thing 15:36:02 <dustins> But yeah, I think this may be specific to the 3PAR driver? 15:36:30 <tbarron> would it make sense to ask if the issue is still reproducible? 15:36:42 <bswartz> I'm not convinced there's any problem outside the 3par driver 15:36:48 <tbarron> or close saying re-open if it's still an issue 15:37:29 <tbarron> we don't have any active HP 3par folks here 15:37:31 <bswartz> The maintainer is Ravichandran Nudurumati <ravichandrann@hpe.com> 15:37:56 <bswartz> No known IRC nick for him 15:38:45 <bswartz> It's up to HPE whether they want to fix this bug or not -- I don't see what we can do without access to their hardware 15:39:17 <dustins> Yeah, I was kinda hoping that we had an HPE representative to provide some feedback 15:39:48 <bswartz> next? 15:39:59 <tbarron> what was the disposition? 15:40:06 <tbarron> assign or close? 15:40:23 <tbarron> we need to clear the backlog, not just talk about the bugs. 15:40:29 <bswartz> I think we leave it unassigned, low importance 15:40:34 <bswartz> We can mark it traiged 15:40:35 <tbarron> -1 15:40:43 <bswartz> triaged even 15:41:04 <tbarron> ok, I'm just looking for ways to make some progress 15:41:31 <bswartz> I don't see how we can close it -- it's likely still an issue that users of the HPE driver should be aware of 15:41:34 * tbarron adds tags 15:42:07 <bswartz> Hopefully if someone users the driver and finds the bug, they can go bother HPE about getting it fixed 15:42:29 <tbarron> tags: drivers, hpe-3par 15:42:34 <xyang> +1 15:42:58 <xyang> we do tags in cinder 15:43:51 <dustins> #link https://bugs.launchpad.net/manila/+bug/1700501 15:43:52 <openstack> Launchpad bug 1700501 in OpenStack Compute (nova) "Insecure rootwrap usage" [Undecided,Incomplete] 15:43:59 <dustins> This is the next one for us 15:44:35 <tbarron> note sdague: "this is too vague to be actiionable" 15:44:58 <tbarron> there's a proposed cross-project initiative for privsep 15:45:00 <bswartz> Wow 15:45:37 <tbarron> we should talk at PTG about whether we should be migrating manila to privsep 15:45:41 <bswartz> Yeah rootwrap was always a pretty hacky approach to fixing security issues 15:45:58 <bswartz> tbarron: Can you add that to the ptg etherpad if its not there already? 15:46:06 <tbarron> bswartz: yes 15:46:08 <bswartz> I don't know the details of privsep 15:46:36 <tbarron> we should talk about what it is and whether / when we want to do it 15:46:37 <bswartz> It's important to note that the security issues addressed by rootwrap are for the most part theoretical issues 15:47:14 <tbarron> yeah, I'm not sayiing that we shouldn't fix specific rootwrap issues (like too broad commands specified, etc.) just 15:47:19 <tbarron> that this bug is open ended 15:47:29 <bswartz> Rootwrap exists to reduce the chances of an exploit in the API services from turning into a more general exploit of the machine the services run on 15:48:23 <bswartz> Our first line of defense is the API services themselves, and while python is bad at many things, it's pretty good at preventing over-the-network exploits 15:50:20 <bswartz> I'm reading about privsep -- is anyone already familiar with it? 15:50:45 <tbarron> kinda sorta, need to refresh 15:51:07 <tbarron> jungleboyj: cinder has some experience attempting privsep migration, not all positive, right? 15:51:40 <bswartz> From what I see, it's unclear that privsep helps at all 15:51:52 * tbarron will try to see where other projects are on this 15:52:10 <jungleboyj> tbarron: It was complicated but we got there. 15:53:00 <jungleboyj> tbarron: Some of it was learning curve for people on how to use it. 15:53:27 <dustins> jungleboyj: Are there good docs on how to transition to privsep? 15:54:04 <jungleboyj> dustins: Not sure. I think hemna did a lot of the work there. Could check with him. 15:54:20 <tbarron> one thing privsep does is encourage use of libraries rather than shell commands to achieve the needed 15:54:29 <bswartz> I'd like someone to explain how privsep is any better than a properly locked-down rootwrap config 15:54:35 <dustins> Not a bad idea, if it's something that other projects are doing we should leverage their experience and expertise 15:54:53 <bswartz> tbarron: I can see that working for some things 15:55:10 <bswartz> In particular stuff like the "chmod 777 /etc/shadow" exploit listed in the bug 15:55:21 <tbarron> right 15:55:29 <bswartz> But so much of what manila does is invoking binaries that will never have python-library equivalents 15:56:17 <bswartz> Then again, much of that is driver specific... 15:56:42 <tbarron> I've got an AI to see where other projects are on quota replacement before PTG, let me see what I can find out about privsep as well. 15:56:43 <bswartz> Perhaps we could modularize the rootwrap and only enable parts of it related to the specific drivers in use 15:56:48 <tbarron> Did it help? 15:56:52 <tbarron> etc. 15:56:56 <bswartz> okay 15:57:00 <bswartz> We're low on time 15:57:05 <bswartz> dustins: anything else? 15:57:27 <tbarron> mention the default share type one quick 15:57:28 <dustins> #link https://bugs.launchpad.net/manila/+bug/1743472 15:57:29 <openstack> Launchpad bug 1743472 in Manila "Create a default share type for tempest tests" [Undecided,New] - Assigned to Victoria Martinez de la Cruz (vkmc) 15:57:39 <vkmc> o/ 15:57:47 <dustins> vkmc: Take it away! 15:57:47 <vkmc> about to submit a fix for that, I'm running test locally 15:57:55 <vkmc> s/test/tempest tests/g 15:57:57 <bswartz> We discuss this one already didn't we? 15:58:12 <tbarron> bswartz: just noting that there's some action on that front ... 15:58:17 <bswartz> Or am I remembering a conversation in another venue? 15:58:23 <vkmc> plus, we need to add the dependency for https://review.openstack.org/#/c/532713/... it's almost a duplicate 15:58:31 <dustins> bswartz: A related one, https://bugs.launchpad.net/manila/+bug/1743472 15:58:32 <openstack> Launchpad bug 1743472 in Manila "Create a default share type for tempest tests" [Undecided,New] - Assigned to Victoria Martinez de la Cruz (vkmc) 15:58:36 <bswartz> Okay 15:58:48 <vkmc> bswartz, yeah, about two weeks ago 15:59:09 <bswartz> Okay let's make sure to get those patches reviewed before RC1 15:59:19 <bswartz> Thanks vkmc 15:59:22 <vkmc> np 15:59:37 <bswartz> Feel free to ping me for a review when it's ready 15:59:45 <bswartz> That's all we have time for today 15:59:57 * bswartz prays the last few patches merge soon 16:00:04 <bswartz> Thanks all 16:00:04 <vkmc> will do, thx 16:00:13 <bswartz> #endmeeting