I am puzzling how I can run coordination-only processes of FoundationDB and have them geographically distributed and separated from the processes that handle actual data (that is for the first time in the same DC). Documentation said that it is good to have distributed coordinators and multi-DC setup is clearly not required.
Thanks!
If I understand correctly, you simply need to add the extra process to the list of coordinators of the main cluster (via the coordinators command of fdbcli, or manual editing of the fdb.cluster).
The extra node will be recruited to help with coordination but nothing else. You must not change the fdb.cluster of the node itself, or else it will attempt to join the cluster (not just as a coordinator).
I think the extra coordinator may also be part of another cluster (hence the necessity to not change its own fdb.cluster file).
I don’t think that a process could be coordinator for two clusters at the same time (but I may be wrong on that).
In fact a process can be used as coordinator by multiple clusters. But this may make it more challenging to do version upgrades, so make sure you understand what you’re doing first.