#opnfv-armband: armbandMarch25
Meeting started by bobmonkman at 14:14:06 UTC
(full logs).
Meeting summary
-
- Alex Avadanii (Enea) (AlexAvadanii1,
14:14:20)
- latest ISO built from our repos deploys and
passes ostf (self-tests) on non-HA setup, with ODL; HA setup is
currently redeployed, seems to work fine too so far (AlexAvadanii1,
14:15:46)
- Can Enea give an update on B-release
deployment (bobmonkman,
14:15:51)
- we are able to able launch VMs without any
manual intervention (AlexAvadanii1,
14:16:25)
- still refining some UX stuff, like web-browser
VNC, or misleading failing tests (AlexAvadanii1,
14:16:53)
- Joe, please confirm that Brahameputr deploy HA
scenario with SDN works fine (bobmonkman,
14:17:01)
- the HA deploy has ODL, does it not?
(joekidder,
14:17:19)
- there are some expected fails with HA
self-tests, since gic v3 does not yet support migrate (not upstream
in kernel, nor in qemu for now) (AlexAvadanii1,
14:17:25)
- joekidder: yes, HA deploy with ODL, what is
left testing there is ceph (AlexAvadanii1,
14:17:41)
- thanks, alex. (joekidder,
14:17:53)
- wel I think HA is orthogonal to ODL or no
ODL (bobmonkman,
14:18:05)
- team, i believe this means we can declare that
Brahamutra reference stack on ARM officially deploys!!! (bobmonkman,
14:18:52)
- if you were to grab the latest armband git repo
and the latest linux.enea.com packages, you should get a fully
working ISO; some minor patches are still refined, but they do not
prevent a succeful deploy with Fuel (AlexAvadanii1,
14:19:04)
- Testing phase now and congrats to the
team (bobmonkman,
14:19:27)
- yes, we still need to run functest and assess
the results, but the deloy is working. Great milestone (bobmonkman,
14:20:13)
- the rest of the team may not have known this
earlier (bobmonkman,
14:20:34)
- so what does pushing to gerrit mean? Please
elaborate (bobmonkman,
14:21:27)
- Currently we are working on gitlab repository
that I've created. From time to time we're pushing to opnfv gerrit
when we have quasi-stable state (StanKardach,
14:22:15)
- we used an internal repo, which we migrated to
armband repo (gerrit) as squashed patch-sets (AlexAvadanii1,
14:22:25)
- current armband repo is 2-3 commits behind
internal repo, mostly fixing UX issues (AlexAvadanii1,
14:22:47)
- OK, so bringing into an opnfv established
repo...thank you (bobmonkman,
14:23:08)
- yes, please give Stan rights to armband, this
is why gerrit is lagging :) (AlexAvadanii1,
14:23:13)
- bob - do you need me to re-send you and aric
(?) the committer votes? I have them in my inbox (the votes were
all public, so I'm not sure what the issue is). (joekidder,
14:23:58)
- apologies. I have opened to ticket to reflect
the vote, but the project has asked for the email approval trail,
which I will do today, then Stan will be official committer along
with other two (bobmonkman,
14:24:29)
- anyone using linux.enea.com: keep in mind this
is Amazon EC2, if you can use a local mirror of that, we would
greatly appreciate it, thank you (AlexAvadanii1,
14:24:33)
- We'd need official space to host all the
packages that we've prepared. EC2 is not the most cost-effective
way (StanKardach,
14:25:18)
- Can I please get an update on connecting to
Jenkins? How is that process going? (bobmonkman,
14:26:48)
- josep: can you give jenkins update?
(joekidder,
14:28:00)
- bobmonkman: there are two tracks, deploy and
buiild. I have been looking at the different configuration files
required for deployments, but I am not there yet (josep,
14:29:28)
- Josep: So do you have what you need to keep
going or is there a blocker? (bobmonkman,
14:30:54)
- bobmonkman: I'll need help from fuel
(josep,
14:31:49)
- bobmonkman: but I don't see any obvious
blockers (except maybe that the deploy script uses ipmitool, which
won't work great on some targets) (josep,
14:32:34)
- but I have not come that far yet to tell
(josep,
14:33:01)
- OK, Joe I assume can help facilitate that and
then do we have any idea of how long we expect the process to be to
get started tetsing under Jenkins? (bobmonkman,
14:33:03)
- some aarch64 "servers" do not support
out-of-band power control - e.g. APM (AlexAvadanii1,
14:33:14)
- @Alex: so is there a way around such
limiitations as cannot run ipmi on a given box? (bobmonkman,
14:34:18)
- I want to get some sort of estimate as to when
we could be ready with Jenkins, please (bobmonkman,
14:36:05)
- simple workaround is a network power switch, as
that's what most of ipmi is used for (I believe) in this setup - to
reboot machines. (joekidder,
14:36:11)
- Can we get a Functest update as well?
(bobmonkman,
14:36:23)
- bobmonkman: in our case, we might have
something, as we use some powercontrollers that support SNMP, and we
can power-on/off the targets with those, but it's not a universal
solution (AlexAvadanii1,
14:36:56)
- bobmonkman: as far as I've seen, the deploy
script powers on/off the targets, and uses ipmi for that. It also
sets the default boot order this way (josep,
14:37:06)
- for virtual deployments it uses libvirt, though
(there is a class abstracting all this) (josep,
14:37:59)
- , so it appears we wil have to request a change
to the deploy scripts for targets that do not support ipmi
(bobmonkman,
14:38:00)
- this is something we could contribute if
necessary, I believe (josep,
14:38:50)
- there is no workaround for boot order though,
not without adding a serial controller to the setup, which again, we
do have in our lab, but not an universal solution (AlexAvadanii1,
14:38:53)
- ...ok something to note as to how to work with
Octopus team or whomever to decide how to handle power cycle and
boot order more generically (bobmonkman,
14:40:16)
- bobmonkman: hard to give a time estimate for
jenkins, but not by next week (josep,
14:40:48)
- OK, josep....we will revisit in 2 weeks. No
meeting next week due to holiday break. (bobmonkman,
14:41:29)
- ,, Ciprian please use #info so we capture in
minutes bot (bobmonkman,
14:45:58)
- unfortunately I had near zero success in
running on Fuel 8, I had multiple problems in my way (ciprian-barbu,
14:46:16)
- I encountered a few bugs as well, I had to
spent time debugging (ciprian-barbu,
14:46:23)
- I had a problem with the network setup in the
lab, I could only run functest on POD1, which turns out has a
problem with libvirt (ciprian-barbu,
14:46:33)
- I thought the libvirt problem was
solved? (bobmonkman,
14:47:05)
- yesterday Magnus from IT solved the
connectivity problem to POD2, so I managed to actually start working
on that (ciprian-barbu,
14:47:09)
- I am continuing with this, I've been delayed so
far (ciprian-barbu,
14:48:22)
- to be clear, Ciprian, you are seeing failures
on Functest, when running Fuel 8 based deploy, that you were not
seeing before on Fuel 6? (bobmonkman,
14:48:33)
- no, there were a few minor things, keystone
client has been deprecated and that was an easy fix (ciprian-barbu,
14:49:26)
- but in fact I wasn't able to run functest at
all on Fuel 8 (ciprian-barbu,
14:49:48)
- in fact, I tried to run, but I bumped into this
issue that seems like libvirt related (ciprian-barbu,
14:50:52)
- I'm not 100% sure about it, could be something
else (ciprian-barbu,
14:51:04)
- OK, team, lets cme up with a plan for the next
2 weeks to help find out why Functest will not run. This is central
to connecting to Jenkins and being in CI loop (bobmonkman,
14:51:57)
- Joe, can you please work with the team to get
on top of this? (bobmonkman,
14:52:49)
- yep. (joekidder,
14:52:56)
- ACTION: Bob will send
IT the committer vote log and committer drop confirmations
(bobmonkman,
14:54:09)
- ACTION: Joe will work
with the team to get over Functest issues (bobmonkman,
14:54:40)
- Next team meeting will be 2 weeks from today. I
will be out for the holiday week and will check in early in the week
of April 4 (bobmonkman,
14:56:06)
- we would need a bit of help with DHA/DEA
files (StanKardach,
14:56:12)
- we can't seem to find documentation on them and
they seem vital in understanding how the deployment testing
works (StanKardach,
14:56:40)
- @StanK: Could this be an Octopus item?
Fuel? (bobmonkman,
14:57:26)
- I'm almost sure it's not Mirantis related but
rather OPNFV-Fuel related (StanKardach,
14:58:03)
- I'm not sure if it's present in other
installers as I'm not familiar with them (StanKardach,
14:58:24)
- OK, I suggest sending a question to the list
asking for help. (bobmonkman,
14:58:47)
- ACTION: Stan to ask
Fuel team for help on documentation of DHA/DEA procedures
(bobmonkman,
15:00:17)
Meeting ended at 15:02:00 UTC
(full logs).
Action items
- Bob will send IT the committer vote log and committer drop confirmations
- Joe will work with the team to get over Functest issues
- Stan to ask Fuel team for help on documentation of DHA/DEA procedures
People present (lines said)
- bobmonkman (44)
- AlexAvadanii1 (18)
- ciprian-barbu (16)
- joekidder (12)
- josep (12)
- StanKardach (11)
- bin_ (4)
- collabot` (3)
Generated by MeetBot 0.1.4.