ramishra | tkajinam: sorry I missed your message yesterday. I think my intention is to provide a workaround till nova fixes the thing as it's blocking customers. Though re-running stack update would possibly fix the issue in most cases, with complex nested stacks it can be painful and messy I guess. | 08:38 |
---|---|---|
ramishra | yeah, if there are still cases where the workaround won't work, we should probably not do it. | 08:39 |
tkajinam | the problem with the given workaround is that it may block the whole wait loop in rare circumstances. | 09:37 |
tkajinam | I mean, the alternative nova api proposed does not return port id, which is identical, so in a rare situation heat can't identify the exact record and may be confused by a different address record. | 09:39 |
tkajinam | ramishra, because I may not be able to read that jira where the issue is discussed, I don't know how critical the problem is. but I hope that update pattern is rare and the customer agrees with trying stack update as a "workaround"... | 09:42 |
ramishra | you can probably read this jira https://issues.redhat.com/browse/OSPRH-17252 to an extent, I think they're open | 10:42 |
ramishra | not sure I understand what you mean by 'port_id which is identical', you mean a scenario where network+mac are the same for different ports attached to same server? | 10:48 |
ramishra | I never knew there could be a scenario like that. | 10:48 |
tkajinam | ramishra, neutron enforces a uniqueness for only network_id + mac | 11:16 |
tkajinam | and neutron does not enforce uniqueness for network name | 11:18 |
tkajinam | so theoretically a user can attach two ports with the same mac address, from two networks, which have the same name. | 11:19 |
tkajinam | to a single instance | 11:19 |
tkajinam | hm though the bug description indicates this is not something nova allows | 11:22 |
opendevreview | Takashi Kajinami proposed openstack/heat master: Fix check_interface_detach() https://review.opendev.org/c/openstack/heat/+/953219 | 12:06 |
tkajinam | ^^^ I'm not fully convinced that we have to implement this really but I won't block it to help that VIP customer :-P | 12:07 |
opendevreview | Takashi Kajinami proposed openstack/heat master: Fix check_interface_detach() https://review.opendev.org/c/openstack/heat/+/953219 | 14:11 |
-opendevstatus- NOTICE: Gerrit is being restarted to pick up a configuration change. You may notice a short outage. | 17:35 |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!