[Pgpool-general] RE: Bad degeneration when shutting down secondary backend

[Josh Berkus]

Pablo,

> Note: All of this, only happens if the environment has continue "heavy" (3
> tps) transaction load. When testing the same conditions with "light" (< 1
> tps) transaction load, pgpool behavies correctly. ¿May I do some further
> research? ¿Why exactly pgpool detects a database shutdown as data
> inconsistency?.

This is consistent with the issues we were seeing.  For that matter, if you 
are in replication mode, and attempt to load 300MB of data in 3 streams, 
pgPool fails with a "data inconsistency".  This happens even if 
replication_stop_on_mismatch = off.

Clearly there's an issue with the data inconsistency detection code; I should 
dig into it.

-- 
Josh Berkus
Aglio Database Solutions
San Francisco