Discussion about CSI driver update in Kubernetes

[Jan Safranek]

Hi,

on last CSI call I promised I'll schedule a meeting on Wednesday.
Unfortunately today I realized I have another meeting there.

Here are my options this week that have some overlap with PDT:
https://doodle.com/poll/752s8k9i8ec4g2rn, please vote. Earlier is
better, of course :-).

For those who were not on the meeting (e.g. couple of Gluster folks in
To:), I'd like to discuss how to safely update a CSI driver in
Kubernetes without disrupting running applications. See
https://docs.google.com/document/d/1opkOXxikHgfPgIw5ee1OwgTzCSDUi0gqcJZ2K1gfmKo/edit#
for details. Join if you want to help or have an idea how to do it.

Jan

[Jan Safranek]

Voting has ended, we'll meet on Friday 8amPDT. Calendar invitation will follow.

In the end it was either Saad and David wake up early or me and Patrick work until 8:30pm on Friday. I choose earlier time. Sorry guys.

Jan

[David Zhu]

No worries! It is definitely better this way. See you Friday.

Best,

David

--
You received this message because you are subscribed to the Google Groups "kubernetes-sig-storage-wg-csi" group.
To unsubscribe from this group and stop receiving emails from it, send an email to kubernetes-sig-stora...@googlegroups.com.
To post to this group, send email to kubernetes-sig...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/kubernetes-sig-storage-wg-csi/1066474016.62234129.1540408546144.JavaMail.zimbra%40redhat.com.
For more options, visit https://groups.google.com/d/optout.

--

David Zhu | Software Engineer | dy...@google.com | 412-436-6859

[Jan Safranek]

I talked to Hemant about re-using attach limit to block scheduler to put
new pods to a node where a CSI driver is being updated. We agreed that
while it could work, it feels like a hack.

Instead, we found another two solutions. Solution 5 in
https://docs.google.com/document/d/1opkOXxikHgfPgIw5ee1OwgTzCSDUi0gqcJZ2K1gfmKo/edit#
needs two relatively simple admission plugins, the rest is done via node
labels + pv/pod affinity.

Solution 6 is even simpler, but needs a new predicate in scheduler.

I am on PTO for next 10 days, I'll return to it when I am back. I'll try
to talk to scheduler people.