Geo Open Office - discussion about the Databases
Description:
This issue is to organize a deep dive discussion provided by the Geo group by way of Douglas. Please, provide any questions related to Geo and its databases. Any architecture, any question.
We would like to record this discussion and add it to an advanced Geo training module.
Questions:
- Can you explain how the WAL files are used, and do they differ in anyway between Postgres clusters and the tracking database?
-
@bprescott_ Do we change the behaviour of WAL logging when enabling Geo; eg to ensure WAL files are big enough and/or enough are kept if a replica is down for a while? I'm guessing it's not possible to do automatically, is there a way we could provide guidance about sizing WAL on the primary? This might not be the sort of question you wanted .. anyway, some context from the docs:
It doesn’t matter how long the secondary site is disconnected from the primary site as it is able to replay all the events in the correct order and become synchronized with the primary site again.
- @bprescott_ Geo DB replication seems to use the same fundamental PostgreSQL replication capability as Patroni/repmgr. Are there any differences in terms of PostgreSQL settings or behaviour?
- Do you have any useful tips and tricks when troubleshooting Geo databases?
- Q1
- Q2
- etc...
Edited by Bo Carbonell