12:00:22 #startmeeting TripleO Security Squad 12:00:23 Meeting started Wed Aug 22 12:00:22 2018 UTC and is due to finish in 60 minutes. The chair is jaosorior. Information about MeetBot at http://wiki.debian.org/MeetBot. 12:00:24 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 12:00:26 The meeting name has been set to 'tripleo_security_squad' 12:01:00 hey lhinds ! 12:01:05 sshnaidm, it's in the stinkin journal 12:01:29 sshnaidm, https://review.openstack.org/#/c/593716/ 12:01:33 raildo, redrobot, are you around? 12:01:39 jaosorior: hey 12:01:40 o/ 12:02:03 I'll wait a bit to see if more folks log in 12:02:06 Tengu: ^^ 12:02:30 jaosorior: :) 12:04:07 o/ 12:04:11 o/ 12:04:19 Alright, lets start 12:04:29 #topic SELinux for containers 12:05:09 EmilienM: -^ might be interesting for you as well :) 12:05:11 So, Tengu has been involved with work related to moving from docker to podman, and on the way, he started checking out the SELinux integration for containers, a topic which we had in our view 12:05:39 Tengu: IIRC, you started looking at this by running the containerized undercloud with podman, and SELinux enabled, right? 12:06:04 indeed. Following EmilienM first steps in order to get an undercloud running on podman instead of docker. 12:06:29 So, Tengu run into some SELinux issues, which he tracked down in this trello card https://trello.com/c/hNkI15a7/1-selinux-issues 12:06:46 and they're not as much issues as I expected :D 12:06:49 Ruck/Rover https://review.rdoproject.org/r/#/c/15732/ we are updating tempestconf to 2.0.0 in queens if you see any failures related to temepstconfiguraiton let us know 12:06:53 Jiri Tomasek proposed openstack/tripleo-ui master: Fix network lines rendering https://review.openstack.org/594938 12:06:58 indeed, but they are nasty :). 12:07:06 they are 12:07:35 So, whoever is interested in working on this, please contact me and Tengu about it, so we can put you up to speed. It's quite interesting work :D 12:07:51 some of the issues are "normal" and require a specific selinux policy. 12:08:04 but most of them exists only because of bad practices :] 12:08:34 right, we need to divide which of these issues should we rectify ASAP, and which should have an "exception" in the selinux policy (and hopefully get fixed later) 12:09:47 the first of the issues is the way we use docker/podman with puppet in order to generate the configurations; basically we bind-mount the /etc/puppet directory (which has a selinux label of etc_t) into the container to a temporary directory, and then attempt to copy that into the /etc/puppet directory in the container 12:10:07 this is not allowed, as containers only have access to reading and executing stuff with etc_t 12:10:17 URGENT TRIPLEO TASKS NEED ATTENTION 12:10:19 https://bugs.launchpad.net/tripleo/+bug/1786764 12:10:20 https://bugs.launchpad.net/tripleo/+bug/1787910 12:10:20 Launchpad bug 1786764 in tripleo "tripleo-ci-centos-7-scenario000-multinode-oooq-container-updates times out on prepare" [Critical,In progress] - Assigned to Sorin Sbarnea (ssbarnea) 12:10:21 Launchpad bug 1787910 in tripleo "OVB overcloud deploy fails on nova placement errors" [Critical,Triaged] - Assigned to Marios Andreou (marios-b) 12:10:21 creating stuff from it, and writing to it, isn't allowed 12:10:56 with the hopes of following an approach where we'll have an immutable host, we're going with the approach of having dedicated hieradata for the containers 12:10:58 also, the first thing docker-puppet.sh does is an rm -rf /etc/puppet/ssl directory in the container. 12:11:11 second step is to add a file in the /etc/puppet/hieradata directory 12:11:16 both actions are forbidden. 12:11:34 first one can be avoided by copying only wanted files. second one is trickier. 12:13:20 right, though having dedicated hieradata for the containers would solve these issues (I think) since we would then have those files with the needed selinux labels, and copy them as needed, trying to keep docker-puppet.py's functionality 12:13:47 So, either we copy the hieradata to a temp location on the host, and relabel that, or we straight generate the hieradata on a container volume 12:15:06 Anyway, this is the stuff that we've been discussing lately, and we'll keep this trello card updated: https://trello.com/c/hNkI15a7/1-selinux-issues 12:15:16 Any questions/feedback/interest in this topic? 12:16:03 sshnaidm, that patch is failing on file not found 12:16:06 not sure why 12:17:27 weshay, because you use "shell: |" and need to use "shell: >" 12:17:49 bah.. 12:18:04 #topic Secret Management update 12:18:04 Sagi Shnaidman proposed openstack/ansible-role-tripleo-modify-image master: log modify image to a log file for humans https://review.openstack.org/593716 12:18:17 redrobot, moguimar, raildo: anything you wanna bring up on this topic? 12:18:18 weshay, ^^ 12:18:20 o/ 12:18:27 started working on the castellan drive 12:18:41 found out today that the castellan-vault tests are not working 12:18:46 already diagnosed the cause 12:19:21 from vault 0.10.0 forward there is a change in the API 12:19:22 jaosorior, not from my side 12:19:23 Sagi Shnaidman proposed openstack/ansible-role-tripleo-modify-image master: log modify image to a log file for humans https://review.openstack.org/593716 12:19:50 so castellan fails to talk to a vault server >= 0.10.0 12:19:51 moguimar: so, is the castellan driver broken as well? 12:20:04 yep, I filed a bug on launchpad 12:20:08 crap 12:20:11 thanks for filing it 12:20:14 #link https://bugs.launchpad.net/castellan/+bug/1788375 12:20:14 Launchpad bug 1788375 in castellan "API changes in vault 0.10.0 causes test to fail." [Undecided,New] 12:20:28 Sagi Shnaidman proposed openstack/ansible-role-tripleo-modify-image master: log modify image to a log file for humans https://review.openstack.org/593716 12:20:29 but the fix is quite simple 12:20:38 I already have a fix in progress 12:21:00 tests passing and all, just need to make it backward compatible with vault < 0.10.0 12:21:13 ade_lee: are we supposed to have access to castellan's launchpad? 12:21:37 moguimar: when you have a fix let me know. thanks for working on this. 12:21:53 I've analyzed the wireshark logs and the vault client itself does some http request to fetch API version 12:22:23 moguimar: right, it does discovery. Maybe we can do that upon first interaction and cache the result. 12:22:24 so I'll add the same behaviour in the castellan driver 12:22:32 yep 12:22:39 awesome 12:22:47 moguimar: thanks for this work 12:22:54 o/ 12:24:49 #topic Any other business 12:24:55 Anything else folks want to bring up to the meeting? 12:27:00 John Trowbridge proposed openstack/tripleo-quickstart-extras master: WIP: Update default for THT resource registry https://review.openstack.org/594944 12:28:26 Alright folks! thanks for joining! 12:28:31 o/ 12:28:36 thanks jaosorior 12:28:37 Just a reminder, the security squad meeting is now every two weeks 12:28:45 so, talk to you here in two weeks! 12:28:48 #endmeeting