Looking at Tasmania on the maps / codfw cluster, it looks like we did not regenerate all tiles after the T159631 incident. This is now in progress.

It also looks like the rendering is slightly different, with different location names being shown. For example, compare this production tile (eqiad) with its codfw equivalent (require a SSH tunnel to one of the codfw maps server). You can see that "Hamilton" is shown on the codfw cluster but not on eqiad. This could cause minor issue for users switching between DC.

Unless you take special measures two tile servers with the same style and data may render labels differently. Generally this is caused by queries which don't fully order their results and is not generally regarded as a problem, so long as the ordering from one server is consistent.

Moving off the sprint board - the Discovery team won't be able to do this work at this time.

@Pnorman could you have a look at the codfw servers and see if we are ready to move on this?

For reference, the puppet change to do: https://gerrit.wikimedia.org/r/#/c/345591/

I used a SSH tunnel to check maps2001.codfw.wmnet and it's serving tiles fine. One problem I noticed is that it is at least two months out of date on what it renders. The database is up to date, so this is proobably from T175123: tileshell does not honor redis configuration in /etc/tilerator/config.yaml.

This is also present on production, so it's no barrier to going active/active.

Change 379530 had a related patch set uploaded (by Gehel; owner: Gehel):
[operations/puppet@production] maps: active/active public interface

https://gerrit.wikimedia.org/r/379530

We are ready to make maps active / active. Patch https://gerrit.wikimedia.org/r/#/c/379530/ is ready to be merged, but I'll let the traffic team (@ema / @BBlack) merge it, they understand the traffic side of it much better than I do!

Change 379530 merged by BBlack:
[operations/puppet@production] maps: active/active public interface

https://gerrit.wikimedia.org/r/379530