[08:22:12] elukey: Hi! btullis and myself are working on bringing cephfs volumes to the dse-k8s-eqiad cluster, which would require pulling in the ceph-csi-cephfs upstream chart. Its structure is _extremely_ similar to the ceph-csi-rbd chart that you already evaluated. Would you be willing to perform a similar evaluation for this one as well? If so, all the [08:22:12] necessary details are in https://wikitech.wikimedia.org/w/index.php?title=Helm%2FUpstream_Charts%2Fceph-csi-cephfs. Thank you kindly [08:23:57] brouberol: of course! Do you have a timeline? I am a bit busy this week, but towards its end I may find some time [08:24:26] towards the end of the week would be perfect! [08:25:11] The charts are extremely similar in nature (the deployment command and args are the main difference) and we have the same set of tweaks we apply on top of it, so you should feel familiar [08:25:20] *it should feel familiar [08:34:12] I've added you as reviewer on the stack patch. Thanks agains! [08:34:26] *again. [09:00:21] elukey: Also adding my thanks in advance. If you have any questions about the setup, please feel free to ask on the patch or direct. These slides might be useful as a general background info about this Ceph cluster: https://docs.google.com/presentation/d/1a1KraypUkZcP-OdwCCqVslEGqetN_pa0qZ9pryYP2rQ/edit [09:01:51] ack thanks! [09:40:32] cephs? years ago, that used to be a nightmare [09:40:59] and was marked for years as not stable (in contrast to the rest of the components which were marked stable) [09:41:38] I sure hope that is no longer the case. [09:46:43] ah it still has experimental features I see https://docs.ceph.com/en/mimic/cephfs/experimental-features/ [09:53:14] This is the current list, since we use the 'reef' release: https://docs.ceph.com/en/reef/cephfs/experimental-features/ [09:55:26] We're not planning on using any of those experimental features. Our use case is for persistent volumes that support the ReadWriteMany (RWX) access mode: https://docs.ceph.com/en/reef/cephfs/experimental-features/ [09:58:02] But we welcome your sharing of your concern. :-) [09:58:36] I pasted the same link twice: https://kubernetes.io/docs/concepts/storage/persistent-volumes/#access-modes [11:15:30] ah, thanks for the correction, I didn't know about the diff mimic <-> reef and that we were targeting reef. Yeah, it was always that aspect (RWX) that was problematic. Anyway, you know better your needs and what you are targeting. [21:37:08] hi k8s friends. I would like to upgrade CoreDNS across all our clusters (sigh). the reason is at https://phabricator.wikimedia.org/T376762 [21:37:23] I'm happy to open another task to discuss/plan the upgrade itself if we want to separate that out [21:38:30] please let me know about any thoughts or concerns