Hi,
On 31 Oct 2019, at 23:03, Matt Riedemann <mriedemos@gmail.com> wrote:
On 10/31/2019 4:15 PM, Slawek Kaplonski wrote:
3. CirrOS guest SSH issues
There are several (some might be duplicates):
http://status.openstack.org/elastic-recheck/index.html#1848078 This one is I think the same as we have reported in https://bugs.launchpad.net/neutron/+bug/1850557 Basically we noticed issues with dhcp after resize/migration/shelve of instance but I didn't have time to investigate it yet.
Hmm, https://review.opendev.org/#/c/670591/ is new to devstack in Train and was backported to stable/stein. I wonder if that is too aggressive and is causing issues with operations where the guest is stopped and started, though for resize/migrate/shelve/unshelve the guest is destroyed on one host and re-spawned on another, so I would think that having a graceful shutdown for the guest wouldn't matter in those cases, unless it has to do with leaving the guest "dirty" somehow before transferring the root disk / creating a snapshot (in the case of shelve). Maybe we should bump that back up to 10 seconds?
I finally spent some time on investigating this issue. I was able to reproduce it locally and I found that the problem is in openvswitch firewall. All is described in [1]. I just pushed patch which should fix this. It’s in [2]. [1] https://bugs.launchpad.net/neutron/+bug/1850557 [2] https://review.opendev.org/696794
--
Thanks,
Matt
— Slawek Kaplonski Senior software engineer Red Hat