They should probably be skipped as we have two masters being written (one per DC) and they need to remain like that - it doesn't matter which DC is active.

@Marostegui Just to clarify and avoid confusion, are you suggesting to remove x2 entirely from the spicerack module and not consider it a core section at all?

FYI the current implementation of the mysql_legacy spicerack module has x2 as part of the CORE_SECTIONS constant and then has an ACTIVE_ACTIVE_SECTIONS constant with just x2 in it to exclude it in some cases.
But the problem is that the get_core_dbs() method assumes that all DBs are part of the A:db-core cumin alias and that's not the case of x2's hosts.
So basically the current implementation of get_core_dbs() is broken with some parameter combination and we need to fix it one way or the other.

[EDIT]
Removing it would mean that it would not be possible to run for x2 methods like check_core_masters_in_sync(), get_core_masters_heartbeats() and check_core_masters_heartbeats()

@Volans I am unsure. How do we treat parsercache at the moment?
x2 is special in the sense that it does have slaves but they're not used (so lag doesn't matter if there any - although it is unlikely there will be any if they are operating normally). It is also special because both of them are actively being written in each DC. So there is no need to go RO for them if we switch DC.
Finally, we don't care about data consistency on those hosts, we are fine if we skip transactions (like we do with parsercache) to a certain extent.

Not sure if this actually helps or not :)

In T329533#8613691, @Marostegui wrote:

@Volans I am unsure. How do we treat parsercache at the moment?

Parsercache is totally ignored and not considered by the module as core sections.

x2 is special in the sense that it does have slaves but they're not used (so lag doesn't matter if there any - although it is unlikely there will be any if they are operating normally). It is also special because both of them are actively being written in each DC. So there is no need to go RO for them if we switch DC.
Finally, we don't care about data consistency on those hosts, we are fine if we skip transactions (like we do with parsercache) to a certain extent.

Not sure if this actually helps or not :)

Given the above I'd say to remove the whole logic about x2 from the module and ignore them completely. If there is nothing to be done and nothing to be checked not worth the added complexity to keep considering them IMHO.

In T329533#8613695, @Volans wrote:

In T329533#8613691, @Marostegui wrote:

@Volans I am unsure. How do we treat parsercache at the moment?

Parsercache is totally ignored and not considered by the module as core sections.

x2 is special in the sense that it does have slaves but they're not used (so lag doesn't matter if there any - although it is unlikely there will be any if they are operating normally). It is also special because both of them are actively being written in each DC. So there is no need to go RO for them if we switch DC.
Finally, we don't care about data consistency on those hosts, we are fine if we skip transactions (like we do with parsercache) to a certain extent.

Not sure if this actually helps or not :)

Given the above I'd say to remove the whole logic about x2 from the module and ignore them completely. If there is nothing to be done and nothing to be checked not worth the added complexity to keep considering them IMHO.

Let's do that then!

Change 889490 had a related patch set uploaded (by Clément Goubert; author: Clément Goubert):

[operations/software/spicerack@master] mysql_legacy: remove x2 handling logic

https://gerrit.wikimedia.org/r/889490

The above patch removes x2 from the core databases, and removes the now unused ACTIVE_ACTIVE_SECTIONS constant while keeping the exclusion logic if we ever need it.

Can I get a firm answer from Data-Persistence that this is the way we want to go before merging and having @Volans do a spicerack release please?

Yes, that's the way we should do it given Manuel's comment above and my basic understanding of how x2 works.

Change 889777 had a related patch set uploaded (by Clément Goubert; author: Clément Goubert):

[operations/cookbooks@master] sre.switchdc.mediawiki: Remove ACTIVE_ACTIVE_SECTIONS

https://gerrit.wikimedia.org/r/889777

Thanks @Ladsgroup once the spicerack release is done I'll test the cookbook patch and we should be good on that front.

Change 889490 merged by jenkins-bot:

[operations/software/spicerack@master] mysql_legacy: remove x2 handling logic

https://gerrit.wikimedia.org/r/889490

Change 889777 merged by jenkins-bot:

[operations/cookbooks@master] sre.switchdc.mediawiki: Remove ACTIVE_ACTIVE_SECTIONS

https://gerrit.wikimedia.org/r/889777