=================== #opnfv-cperf: cperf =================== Meeting started by dfarrell07 at 19:09:50 UTC. The full logs are available at http://ircbot.wl.linuxfoundation.org/meetings/opnfv-cperf/2017/opnfv-cperf.2017-01-12-19.09.log.html . Meeting summary --------------- * Jack is reporting that Intel Pod 2 is back up and ready for our use (dfarrell07, 19:10:21) * Jumphost now has same hardware as other nodes (relevant to NSTAT running on jumphost discussion) (dfarrell07, 19:10:47) * Jamo has VPN creds, Nikos and team have open JIRA with Jack to get creds, need to provide GPG info (dfarrell07, 19:11:20) * dfarrell07 has been talking with Veryx folks about PktBlaster (dfarrell07, 19:13:11) * Veryx has Dockerfile with PktBlaster install, plus some helper shell scripts and config files (dfarrell07, 19:13:29) * They are currently on some FTP server, need to get Veryx to submit via Gerrit so can do proper code review (dfarrell07, 19:13:54) * dfarrell07 gave basic docs+pointers to Veryx folks on Git/Gerrit tooling, described where to put their changes in cperf's repo (dfarrell07, 19:14:30) * Jamo has POD hooked back up to jenkins, can start job again, but failed (dfarrell07, 19:19:48) * One of our jobs broke because apex dropped L2 scen, Jamo fixed that (dfarrell07, 19:23:48) * Marcus et al have been working on control container this week (have compute working) (dfarrell07, 19:24:22) * Marcus is working with ansible logic for large set of OpenStack nodes, compute, and having opendaylight run it (dfarrell07, 19:26:57) * Current Rally tests don't spin up VMs on compute nodes, which is where a lot of ODL's work comes in, so need to expand automated tests to do that (have done manually) (dfarrell07, 19:28:15) * Discussion about HPE person hitting OOM doing NetVirt scale, Marcus seems to not be hitting it, discussion about details follow (dfarrell07, 19:29:06) * Marcus et al are mostly working on deployment logic so far, so maybe going to hit OOM issues later (dfarrell07, 19:29:30) * Thread is on controller-dev mailing list (OOM at 1k nodes) (dfarrell07, 19:29:44) * Marcus maxed out RAM on systems they are using (beefy) (dfarrell07, 19:30:02) * Marcus has been working on Kolla Ansible stuff (dfarrell07, 19:30:20) * Marcus asks about config knobs ODL deploys need (dfarrell07, 19:32:52) * LINK: https://github.com/dfarrell07/puppet-opendaylight/blob/master/manifests/config.pp Config knobs used in puppet, as required by OPNFV (dfarrell07, 19:35:37) * Discussion about not using devstack to build these nodes, how to do better deployments (that don't require a stack each time) (dfarrell07, 19:38:17) * One option is to pull down pre-built Apex images, do any virt customize steps we need (to upgrade ODL version say), would already have all O/S services installed and config'd and running (dfarrell07, 19:39:13) Meeting ended at 19:40:20 UTC. People present (lines said) --------------------------- * dfarrell07 (23) * collabot (3) Generated by `MeetBot`_ 0.1.4