I posted this in the pgpool mailing list and so far has not recvd any response. Looks like that mailing list is hardly
active.
=========================================
I am researching pgpool technology. Let us say we want to use it as an alternative to streaming replication of PG.
Apparentlyin replication mode, pgpool can replicate to all participating nodes, which presumably will be in raw mode
(nonreplicating mode).
My questions:
1 - Do we have a choice of limiting number of nodes for sync writes (and remaining one in async), so that overall
performanceimpact is limited.
2 - If a node is down, does pgpool has the ability to resync from the last known transaction that was committed. How
doesit do it? Does it maintain
a state information or save all transactions that need to be replayed later on.
3 - Is app-stickiness possible. By app-stickiness I mean, a given app or session always goes to the same node. This
helpsin avoiding stale reads.
4- Is the failover transparent to the client. Meaning, if a node goes down, do clients have to use a new connection
string.
5 - How about high availability of pgpool itself. Can we have multiple pgpools?
If there is a documentation to answer these questions, please let me know.
thanks.