Thursday, 2025-12-04

opendevreviewmelanie witt proposed openstack/devstack-plugin-ceph master: DNM try without debian 12  https://review.opendev.org/c/openstack/devstack-plugin-ceph/+/96952300:22
opendevreviewInyong Hong proposed openstack/manila master: NetApp: Update created_at during manage operations  https://review.opendev.org/c/openstack/manila/+/96954601:11
opendevreviewInyong Hong proposed openstack/manila master: NetApp: Update created_at during manage operations  https://review.opendev.org/c/openstack/manila/+/96954601:12
opendevreviewInyong Hong proposed openstack/manila master: NetApp: Update created_at during manage operations  https://review.opendev.org/c/openstack/manila/+/96954601:12
opendevreviewInyong Hong proposed openstack/manila master: NetApp: Update created_at during manage operations  https://review.opendev.org/c/openstack/manila/+/96954601:13
opendevreviewInyong Hong proposed openstack/manila master: NetApp: Update created_at during manage operations  https://review.opendev.org/c/openstack/manila/+/96954601:21
hong-p[m]hi gouthamr carloss During the last PTG we discussed improving the created_at value for share and snapshot manage so that it reflects the actual creation timestamp in the storage backend instead of the time when Manila starts managing the resource. I have now created the corresponding Blueprint and a Change Request (CR) for this change. Could you please take a look and review? https://review.opendev.org/c/openstack/manila/+/96954601:34
opendevreviewkiran pawar proposed openstack/manila-specs master: Add spec for qos types  https://review.opendev.org/c/openstack/manila-specs/+/96270608:28
opendevreviewMerged openstack/manila master: Drop eventlet usage in ssh pools  https://review.opendev.org/c/openstack/manila/+/96023109:34
carlosshey hong-p[m] - thanks for creating the blueprint and proposing the change10:11
carlossI'll take a look at both10:11
opendevreviewCarlos Eduardo proposed openstack/manila master: [WIP] Fix rally jobs  https://review.opendev.org/c/openstack/manila/+/96959410:45
opendevreviewCarlos Eduardo proposed openstack/manila master: [WIP] Fix rally jobs  https://review.opendev.org/c/openstack/manila/+/96959410:51
opendevreviewCarlos Eduardo proposed openstack/manila master: [WIP] Fix rally jobs  https://review.opendev.org/c/openstack/manila/+/96959411:22
opendevreviewCarlos Eduardo proposed openstack/manila master: Fix rally jobs  https://review.opendev.org/c/openstack/manila/+/96959412:06
opendevreviewCarlos Eduardo proposed openstack/manila master: Fix rally jobs  https://review.opendev.org/c/openstack/manila/+/96959412:37
opendevreviewkiran pawar proposed openstack/manila master: Add support for QoS type and specs  https://review.opendev.org/c/openstack/manila/+/96782214:52
opendevreviewGoutham Pacha Ravi proposed openstack/manila stable/2025.2: Fix shares getting stuck in ensuring status  https://review.opendev.org/c/openstack/manila/+/96979515:29
gouthamrKumar_T: from my understanding, NetApp supports migrating share servers that have replicated shares.. 16:07
gouthamrthe doc probably needs to call out there's a driver-reported attribute called: "share_replicas_migration_support" that allows/denies server migrations with replicated shares16:07
Kumar_TOk. That guideline can be overridden by flag16:08
kpdevyes, we support migration with replicas and also removed all blocking code from API in our downstream repo16:08
Kumar_TYes16:08
gouthamrspecifically, if admins are concerned that they'd not be able to migrate share servers when replicated shares show up, they should be using that attribute as a share type extra-spec 16:08
gouthamrthat way they can prevent creation of replicated shares on DHSS=True backends where servers can't be migrated with replicas16:09
Kumar_TKpdev you recall which version?16:09
kpdevi can not recollect version. may be git log . will help to see when migration with replica support was added16:10
gouthamrhttps://bugs.launchpad.net/manila/+bug/205278516:12
gouthamrhttps://review.opendev.org/c/openstack/manila/+/93390216:12
gouthamrThis issue was fixed in the openstack/manila 20.0.0.0rc1 Epoxy release candidate16:12
kpdevI had worked on this.. sorry I was not able to remeber it16:13
kpdevanyone has any questions on QoS spec, please raise. I will be here next 20 minuts.16:15
Kumar_Tkpdev16:17
Kumar_TQoS spec feature is applicable for both DHSS mode ? Or only for DHSS true case?16:18
Kumar_TNext question: There will be drivers like NetApp who supports qos creation/usage using share extra specs already.16:19
Kumar_Thttps://netapp-openstack-dev.github.io/openstack-docs/draft/manila/deployment_choice/section_creating_service_catalog.html#table-6-1216:19
kpdevfor both, But please check internally16:19
kpdevAnoop has commented to keep this for DHSS=True and use legacy way of adaptive_qos i.e. share_type extra-spec for DHSS=False16:20
kpdev>Next question: There will be drivers like NetApp who supports qos creation/usage using share extra specs already.16:20
kpdevyes, we know this16:20
Kumar_TIf user pass both in extra specs what is the outcome? Error? or some precedence?16:20
Kumar_TSure lets discuss internally on driver behavior16:21
kpdevThe create API is ADMIN only. so it is assumed that admin will configure share_type and qos_type correctly.16:21
* carloss following the conversation, wanted to share a suggestion16:22
carloss> But please check internally16:23
carlosskpdev internally you mean internal chats?16:23
kpdevThe existing deployment can keep using specifying spec for fixed policy i.e. share type extra specs. and adaptive the same way for dhss=false as it takes only name16:23
kpdevfor dhss=true we need all params i.e. using qos_type.16:24
kpdevinternally means within Netapp team, I had couple of sync up calls with Netapp for Qos since last 2 months and after discussion and use case, the spec is created.16:24
kpdevSo it was more suggestion from Netapp only to keep existing way for dhss=False and use new way for DHSS=True. In either case, Admin should make the correct configuration of qos_type and its mention in share-type extra-specs16:25
carlosskpdev ack, thanks... this is an interesting discussion and is nice context for upstream as well. Is it okay if I ask you guys to please use #openstack-manila for discussions around proposed specs/functionalities? I think it also helps everyone upstream16:26
carlossof course we should not drop customer names16:27
carlossbut a question like this: QoS spec feature is applicable for both DHSS mode ? Or only for DHSS true case?16:27
carlossis definitely good context for upstream16:27
kpdevThe spec ideally can be used for both DHSS=False and True. For SAP use case only DHSS=True as that is only scenario we support.16:28
carlossdriver behavior is also something nice to keep the discussion upstream :)16:29
carlosssorry, trying to keep these discussions here so that you guys don't need to answer the same questions if someone upstream asks16:29
carlossalso helps on getting feedback earlier16:29
kpdevwe dont have deployment with fixed QoS and also no deployment with DHSS=False. So adaptive_qos and dhss=true was use case behind this. But since its way of specifying policy, this can be applied to fixed as well.16:30
kpdevsince there will be some deployments with netapp having adaptive qos and dhss=false, Anoop asked to keep this support as it is.16:32
carloss> The create API is ADMIN only. so it is assumed that admin will configure share_type and qos_type correctly.16:32
carlossmakes sense and that's a good question, Kumar_T - should we do checks to prevent this from happening though?16:32
kpdevthe share type and qos type are passed to driver and driver can invalidate the share creation if both things exist w.r.t given share create request . Is it ? @carloss ?16:35
carlossyeah, or even in the api kpdev: that conflict can be identified earlier than in the driver16:35
carlossif both are provided and both have conflicting specs, then we'd know earlier16:36
kpdevin api layer then it will be driver specific check. are you ok with it ?16:36
carlosskpdev: not driver specific check, extra spec check16:36
carlossdifferent things16:36
kpdevwhich extra spec check ?16:37
kpdevif default_qos_type and netapp:max_iops exist in extra-spec ?16:38
carlosswhat I mean is: qos is currently set through extra specs, right? qos type will also have key value specs, right?16:38
carlossif qos type has max_iops and share type has max_iops, what happens?16:39
carlossor am I misunderstanding how we'll set the qos specs to the qos type?16:39
kpdevshare_type wont have max_iops. this will be netapp:max_iops and so the check will be driver specific16:40
kpdevso let driver handle mismatch16:40
carlossyep, that works16:40
kpdevAck16:40
kpdevLeaving, thanks !16:49
opendevreviewGoutham Pacha Ravi proposed openstack/manila stable/2025.1: Fix shares getting stuck in ensuring status  https://review.opendev.org/c/openstack/manila/+/96982717:57
opendevreviewMerged openstack/manila stable/2025.2: Fix shares getting stuck in ensuring status  https://review.opendev.org/c/openstack/manila/+/96979518:42
opendevreviewGoutham Pacha Ravi proposed openstack/python-manilaclient master: Add missing api_versions decorator for _update_share_group  https://review.opendev.org/c/openstack/python-manilaclient/+/96843618:43
opendevreviewMerged openstack/manila master: Fix rally jobs  https://review.opendev.org/c/openstack/manila/+/96959419:21

Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!