@JMeybohm as discussed today:

• Redis hosts upgrade should be done at the same time as HW refreshes tasks T418918, T418261 and T418924 - @Clement_Goubert to confirm
• Parsoid test host should go away after T386246 - @jijiki to confirm
• Kafka-main hosts should be upgraded at the same time as cluster upgrade - @JMeybohm @jasmine_

So we'd be left with:

• configcluster (conf*) hosts
• deploy1003
• deploy2002

In T419212#11681013, @MLechvien-WMF wrote:

• deploy1003
• deploy2002

I think the deploy hosts might be blocked by T419049: Upgrade the MediaWiki servers to ICU 72 ☂️

In T419212#11681013, @MLechvien-WMF wrote:

@JMeybohm as discussed today:

• Redis hosts upgrade should be done at the same time as HW refreshes tasks T418918, T418261 and T418924 - @Clement_Goubert to confirm

Yes, we need to confirm compatibility between redis versions in Bullseye and Trixie and come up with a plan, but that's the goal.

In T419212#11688136, @Clement_Goubert wrote:

In T419212#11681013, @MLechvien-WMF wrote:

@JMeybohm as discussed today:

• Redis hosts upgrade should be done at the same time as HW refreshes tasks T418918, T418261 and T418924 - @Clement_Goubert to confirm

Yes, we need to confirm compatibility between redis versions in Bullseye and Trixie and come up with a plan, but that's the goal.

IIRC Redis does not support skipping a major version, so we probably have to update to bullseye (redis 7) instead of trixie (redis 8).

For conf* hosts, I'd propose we start in codfw, by way of T418915: conf200[7-9] implementation tracking (i.e., conf200[4-6] go way). Once we're happy with that, we can proceed with migrating eqiad (a more complicated procedure, as it involves a proper etcd primary cluster switchover to codfw).

In T419212#11688224, @JMeybohm wrote:

In T419212#11688136, @Clement_Goubert wrote:

In T419212#11681013, @MLechvien-WMF wrote:

@JMeybohm as discussed today:

• Redis hosts upgrade should be done at the same time as HW refreshes tasks T418918, T418261 and T418924 - @Clement_Goubert to confirm

Yes, we need to confirm compatibility between redis versions in Bullseye and Trixie and come up with a plan, but that's the goal.

IIRC Redis does not support skipping a major version, so we probably have to update to bullseye (redis 7) instead of trixie (redis 8).

Filed T419976 as a potential way forward, skipping upgrading to redis 7.

My impression is that @Scott_French is taking the operations touching conf* hosts this quarter.

And for the deploy* hosts, given the dependency on ICU upgrade @Raine can you plan to do them?

@MLechvien-WMF - Yes, exactly. My plan is for the new conf* hosts being racked in codfw to run Trixie from day 1, which will pave the way for upgrading eqiad to Trixie as well (T419212#11703456). The actual process is going to be rather involved, and I'll need to spend a bit of time to iron out the procedure first.

In T419212#11809310, @MLechvien-WMF wrote:

And for the deploy* hosts, given the dependency on ICU upgrade @Raine can you plan to do them?

Yes, I'll take care of them, marking on task.

Change #1283988 had a related patch set uploaded (by Jasmine; author: Jasmine):

[operations/puppet@production] kafka-main: set codfw brokers inter-broker protocol to 3.7

https://gerrit.wikimedia.org/r/1283988

Chatting with @Raine we'll upgrade the deploy role to Bookworm (tracked in T418262: deploy2003 implementation tracking) this quarter. Upgrade to trixie is not high priority for this quarter