I'm reading up on HA configurations and I have tons of questions...
(1) Is it correct that there are two election processes: one that selects a node that will do the monitoring work and one that selects the node that will maintain the IDO database connection? (2) These two jobs talk to each other and they can run on the same machine, but they could also end up on different machines.
(3) What happens when the database goes down for some time? I understand that this may trigger a new election process, but what if the database is unavailable for a lengthy period of time? (4) Will we see one election after the other? (5) Will the data to be written into the database be buffered somewhere until the DB is on-line again? (6) How does this buffer work and what are the limits?
I understand that if all IdoPgsqlConnection objects have enable_ha=false, all nodes will maintain an IDO connection and write to the database. Let's assume that each node has its own database. (7) Does this mean that, at least as long as both machines are working flawlessly, I will end up with two identical copies of the data?
Now I would want some database redundancy as well. (8) Does the following setup make sense: Two machines, each running an icinga2 instance as well as a PosgreSQL server. The two icinga2 instances are in the same zone, but each one is configured to use its own local DB (enable_ha=true). A bi-directional replication between the two databases will make sure that records written into each database will be copied to the other one. icingaweb2 could then connect to either database.
(1) Is it correct that there are two election processes: one that selects a node that will do the monitoring work and one that selects the node that will maintain the IDO database connection? (2) These two jobs talk to each other and they can run on the same machine, but they could also end up on different machines.
(3) What happens when the database goes down for some time? I understand that this may trigger a new election process, but what if the database is unavailable for a lengthy period of time? (4) Will we see one election after the other? (5) Will the data to be written into the database be buffered somewhere until the DB is on-line again? (6) How does this buffer work and what are the limits?
I understand that if all IdoPgsqlConnection objects have enable_ha=false, all nodes will maintain an IDO connection and write to the database. Let's assume that each node has its own database. (7) Does this mean that, at least as long as both machines are working flawlessly, I will end up with two identical copies of the data?
Now I would want some database redundancy as well. (8) Does the following setup make sense: Two machines, each running an icinga2 instance as well as a PosgreSQL server. The two icinga2 instances are in the same zone, but each one is configured to use its own local DB (enable_ha=true). A bi-directional replication between the two databases will make sure that records written into each database will be copied to the other one. icingaweb2 could then connect to either database.