| opendevreview | Eunkyung Kim proposed openstack/manila master: Allow using mount_point_name when managing share https://review.opendev.org/c/openstack/manila/+/970786 | 01:33 |
|---|---|---|
| opendevreview | Eunkyung Kim proposed openstack/manila master: [NetApp] Fix mount format while managing share https://review.opendev.org/c/openstack/manila/+/975372 | 01:33 |
| opendevreview | Merged openstack/manila-specs master: Drop obsolete sphinx extensions https://review.opendev.org/c/openstack/manila-specs/+/976984 | 05:59 |
| opendevreview | Prasant Padhi proposed openstack/manila master: PowerScale documentation update https://review.opendev.org/c/openstack/manila/+/975155 | 06:26 |
| opendevreview | Manideep proposed openstack/manila master: Implementation of NAE support for NetApp Driver https://review.opendev.org/c/openstack/manila/+/977290 | 06:55 |
| *** tkajinam_ is now known as tkajinam | 10:13 | |
| opendevreview | Kumar Tadepalli proposed openstack/manila master: NetApp - Add support for Synchronous replication for share https://review.opendev.org/c/openstack/manila/+/973578 | 11:46 |
| manideep | https://www.irccloud.com/pastebin/FUOy45gS | 12:49 |
| manideep | Hello Team, | 12:50 |
| manideep | 12:50 | |
| manideep | I am Manideep from the NetApp team. | 12:50 |
| manideep | This is on a recent Zuul issue. We have observed that Ubuntu Jammy support was removed from OpenStack Devstack repository on Feb 17th [https://review.opendev.org/c/openstack/devstack/+/976893]. Since then, our NetApp CI has been broken and unable to run tests as we rely on Ubuntu Jammy to install Devstack alongside node pool to run Tempest tests. | 12:50 |
| manideep | 12:50 | |
| manideep | Is anyone else impacted by this change? Any recommended workarounds or plans to reintroduce Jammy support? If anyone is aware on why this change was introduced and can help us with the context it will be great. Thanks for the support. | 12:50 |
| carloss | hey manideep the upstream CI has already migrated to ubuntu 24 (Noble Numbat) | 12:59 |
| carloss | when that happens, unfortunately we stop testing and move on to the next version, meaning the community does not plan on getting back | 12:59 |
| carloss | how's the CI failing? is it okay for you folks to rebuild the new image based on ubuntu 24? | 13:00 |
| opendevreview | kiran pawar proposed openstack/manila master: Add support for QoS type and specs https://review.opendev.org/c/openstack/manila/+/967822 | 13:04 |
| opendevreview | kiran pawar proposed openstack/manila master: Netapp: Add support for QoS type specs for dhss=true https://review.opendev.org/c/openstack/manila/+/973962 | 13:04 |
| manideep | carloss: CI is failing during devstack installation as jammy is now not present in SUPPORTED_DISTROS. > is it okay for you folks to rebuild the new image based on ubuntu 24? We will give it a try, but building new imaged did not go well for us earlier, we were hitting multiple issues. | 13:10 |
| carloss | > carloss: CI is failing during devstack installation as jammy is now not present in SUPPORTED_DISTROS. | 13:11 |
| carloss | ack | 13:11 |
| tkajinam | manideep, as carloss pointed ubuntu jammy was removed from upstream test coverage this cycle, and any jobs should be migrated to noble | 13:38 |
| tkajinam | carloss manideep , this is the devstack change which removed jammy support, just fyi. https://review.opendev.org/c/openstack/devstack/+/976893 | 13:38 |
| tkajinam | oh the link was already posted. ignore it. | 13:38 |
| *** vhari_ is now known as vhari | 14:59 | |
| opendevreview | kiran pawar proposed openstack/manila master: Add support for QoS type and specs https://review.opendev.org/c/openstack/manila/+/967822 | 15:20 |
| opendevreview | kiran pawar proposed openstack/manila master: Netapp: Add support for QoS type specs for dhss=true https://review.opendev.org/c/openstack/manila/+/973962 | 15:20 |
| carloss | Anoop_Shukla: o/ happy to continue the chat here :) | 16:05 |
| Anoop_Shukla | Thanks Carloss. | 16:09 |
| Anoop_Shukla | So today, on Manila when we create the replica we immediately make the replica out_of_sync and wait for the update job to run (in a 5 min cadence) that then marks the replica in_sync based on the storage replication status. | 16:10 |
| Anoop_Shukla | While this works for Async policies, sync policies are expected to have near 0 RTOs. So a 5 min delay in updating the replica status adds to the RTOs. | 16:11 |
| Anoop_Shukla | Customers may not accept that. So what is the guidance here? One way is to mark the replica in_sync immediately (since storage would do the same and we have to test this behaviour) and let the update job make it out_of_sync in case it isn’t in_sync.. | 16:12 |
| Anoop_Shukla | We do not want to add another looping call since we will not have the replica reference to pick on the driver unless it comes from core with a list of out_of_sync replicas so that we can make them in_sync as part of a looping task. | 16:13 |
| Anoop_Shukla | @Kumar | 16:13 |
| Anoop_Shukla | @Kumar_T: can add more when he sees this message | 16:14 |
| gouthamr | you can update the replica_state when returning from the create_replica call in the driver, can’t you? | 16:14 |
| Anoop_Shukla | Yes we can..we were thinking to make an ONTAP ZAPI/REST call with a timeout of say 10-20 with 1-2 seconds loops, and if the replica comes in sync on ONTAP we can mark it in_sync immediately | 16:15 |
| Anoop_Shukla | We can do this only for sync and strict-sync type replication..leave it to be updated by update job for async | 16:17 |
| carloss | Anoop_Shukla: that makes sense to me | 16:19 |
| carloss | you'd be making the share/manager wait on the response, but it's different from making the user wait | 16:20 |
| gouthamr | an alternative would be to mark it in_sync and wait for the periodic update to report errors | 16:20 |
| Anoop_Shukla | Yep..that was the alternative solution that was in my mind | 16:21 |
| carloss | makes sense too + would get rid of the downside I just mentioned (share/manager would be waiting) | 16:21 |
| Anoop_Shukla | But my only concern with that approach is, if ONTAP takes say a little bit more time to bring the mirror into sync state and someone starts a RTO event..based on the share replica state, he may fail mounting | 16:21 |
| gouthamr | isn’t the end of the world, it can be documented behavior.. no one expects to promote a replica as soon as it is created.. | 16:22 |
| gouthamr | except maybe our tempest tests | 16:22 |
| Anoop_Shukla | hmmm | 16:22 |
| Anoop_Shukla | think I got the ideas I needed. We will come back on this once we discuss it internally. Thanks @carloss and @gouthamr. | 16:23 |
| opendevreview | Merged openstack/manila master: tests: Remove tests for v1 API https://review.opendev.org/c/openstack/manila/+/959552 | 18:03 |
| opendevreview | Carlos Eduardo proposed openstack/manila master: Allow using mount_point_name when managing share https://review.opendev.org/c/openstack/manila/+/970786 | 18:27 |
| opendevreview | Carlos Eduardo proposed openstack/manila master: Allow services API to filter by ensuring https://review.opendev.org/c/openstack/manila/+/928081 | 18:27 |
| opendevreview | Carlos Eduardo proposed openstack/python-manilaclient master: Support filtering services by ensuring field https://review.opendev.org/c/openstack/python-manilaclient/+/929327 | 18:47 |
| opendevreview | Merged openstack/manila-tempest-plugin master: Use official service type https://review.opendev.org/c/openstack/manila-tempest-plugin/+/976938 | 19:40 |
| opendevreview | Merged openstack/manila master: api: Remove v1 API https://review.opendev.org/c/openstack/manila/+/959554 | 19:56 |
| opendevreview | Merged openstack/manila master: api: Remove use of mixin classes https://review.opendev.org/c/openstack/manila/+/959555 | 20:23 |
| carloss | ^ another end of an era situation? :) | 20:24 |
| carloss | stephenfin and gouthamr: thank you very much for the effort to remove the manila v1 API | 20:25 |
| gouthamr | haha, indeed.. ty for the reviews! and stephenfin that did all the heavy lifting | 20:34 |
| opendevreview | Goutham Pacha Ravi proposed openstack/manila master: Follow up on API v1 removal https://review.opendev.org/c/openstack/manila/+/977375 | 21:55 |
| opendevreview | Carlos Eduardo proposed openstack/manila master: Allow using mount_point_name when managing share https://review.opendev.org/c/openstack/manila/+/970786 | 22:09 |
| opendevreview | Carlos Eduardo proposed openstack/manila master: Allow using mount_point_name when managing share https://review.opendev.org/c/openstack/manila/+/970786 | 22:22 |
| opendevreview | Eunkyung Kim proposed openstack/manila master: [NetApp] Fix mount format while managing share https://review.opendev.org/c/openstack/manila/+/975372 | 22:23 |
| opendevreview | Merged openstack/python-manilaclient master: Add mount_point_name option for manage shares https://review.opendev.org/c/openstack/python-manilaclient/+/970787 | 22:31 |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!