Resolving this after a successful MediaWiki switchover to codfw and switchback to eqiad using the automation software Switchdc (operations-switchdc on gerrit). The tracking task for improvements is T163363.

Change 351313 merged by Giuseppe Lavagetto:
[operations/puppet@production] cache::text: switch all mediawiki to eqiad

Change 351315 merged by Giuseppe Lavagetto:
[operations/puppet@production] discovery::app_routes: switch mediawiki to eqiad

Change 351539 merged by jenkins-bot:
[operations/mediawiki-config@master] wmf-config: readonly is set in etcd now

Change 351132 merged by jenkins-bot:
[operations/mediawiki-config@master] Enable EtcdConfig in production

Change 351539 had a related patch set uploaded (by Volans; owner: Volans):
[operations/mediawiki-config@master] wmf-config: readonly is set in etcd now

Change 351346 merged by Volans:
[operations/switchdc@master] t05_switch_datacenter: fix typo in DNS checks

Change 351346 had a related patch set uploaded (by Volans; owner: Volans):
[operations/switchdc@master] t05_switch_datacenter: fix typo in DNS checks

Change 351335 merged by Volans:
[operations/switchdc@master] t04: fix title

Change 351335 had a related patch set uploaded (by Volans; owner: Volans):
[operations/switchdc@master] t04: fix title

Change 351315 had a related patch set uploaded (by Volans; owner: Volans):
[operations/puppet@production] discovery::app_routes: switch mediawiki to eqiad

Change 351313 had a related patch set uploaded (by Volans; owner: Volans):
[operations/puppet@production] cache::text: switch all mediawiki to eqiad

thanks @tstarling!

In T156924#3226349, @tstarling wrote:

In T156924#3224786, @Volans wrote:

Is there an easy way I could check which version and/or value of an Etcd-driven MW-config variable is actually loaded/cached by the running application?

I'll add a siteinfo hook or something.

Change 351232 merged by jenkins-bot:
[operations/mediawiki-config@master] Add $wmfMasterDatacenter to meta=siteinfo

Change 351217 merged by jenkins-bot:
[mediawiki/core@wmf/1.29.0-wmf.21] [1.29.0-wmf.21] config: Use less generic cache key, and not fragmented by wiki

Change 351232 had a related patch set uploaded (by Tim Starling; owner: Tim Starling):
[operations/mediawiki-config@master] Add $wmfMasterDatacenter to meta=siteinfo

Change 351204 merged by jenkins-bot:
[mediawiki/core@master] config: Use less generic cache key, and not fragmented by wiki

Change 350969 merged by jenkins-bot:
[mediawiki/core@master] Avoid master queries in loadAndLazyInit() for miser mode

Change 351217 had a related patch set uploaded (by Tim Starling; owner: Krinkle):
[mediawiki/core@wmf/1.29.0-wmf.21] [1.29.0-wmf.21] config: Use less generic cache key, and not fragmented by wiki

Change 351204 had a related patch set uploaded (by Krinkle; owner: Krinkle):
[mediawiki/core@master] config: Use less generic cache key, and not fragmented by wiki

In T156924#3224690, @tstarling wrote:

Did the following testing: [..]

• Confirmed that the APC cache was operational, by sending 100 requests to Special:BlankPage with ab while watching for inbound connections on the etcd server using tcpdump. There were 4 connections in 12 seconds. Connections could be seen coming in pairs with ~10s spacing between pairs.

In T156924#3224786, @Volans wrote:

Is there an easy way I could check which version and/or value of an Etcd-driven MW-config variable is actually loaded/cached by the running application?

Is there an easy way I could check which version and/or value of an Etcd-driven MW-config variable is actually loaded/cached by the running application?

Change 351132 had a related patch set uploaded (by Tim Starling; owner: Tim Starling):
[operations/mediawiki-config@master] Enable EtcdConfig in production

Did the following testing:

Change 347537 merged by jenkins-bot:
[operations/mediawiki-config@master] Use EtcdConfig in beta cluster only

Change 350969 had a related patch set uploaded (by Aaron Schulz; owner: Aaron Schulz):
[mediawiki/core@master] Avoid master queries in loadAndLazyInit() for miser mode

In T156924#3220304, @Volans wrote:

1. Should MW work seamlessly if there is a temporary outage on the etcd service?
   • I think it should and we should use the stale cached data if that happens, assuming that all cached configuration is still valid.

Change 350818 merged by Volans:
[operations/switchdc@master] Mediawiki: update role name for maintenance

Change 350818 had a related patch set uploaded (by Volans; owner: Volans):
[operations/switchdc@master] Mediawiki: update role name for maintenance

Regarding the implementation of the MW configuration, in particular CR https://gerrit.wikimedia.org/r/#/c/347537 (current patchset is #8), I think that we should first agree on the failure model, because I've seen different comments and approaches.

Change 349468 merged by Alexandros Kosiaris:
[operations/puppet@production] profile::conftool::master: make the git root dir a parameter

This is awesome! Thanks @Joe for getting this working on beta.

To summarize, I think it is possible to test EtcdConfig in beta at this point with the limited deployment I made.