[pgpool-general: 5333] Potential Bug: Frequent Unnecessary Degeneration

David O'Mahony david.c.omahony at gmail.com
Thu Feb 16 09:57:23 JST 2017

Hi All,

We're running two nodes using with replication enabled. 

pgpool routinely (every day) performs a failover with the following statements appearing the in log:

2017-02-15 13:16:01: pid 16190: WARNING:  write on backend 1 failed with error :"Success"
2017-02-15 13:16:01: pid 16190: DETAIL:  while trying to write data from offset: 0 wlen: 5

The source code (pool_stream.c:pool_flush_it()) seems to consider a write of 0 bytes an error.  Also the "Success" in the log statement indicates that errno was not set.  

There never appears to be any actual problem with the node.

Should pgpool trigger a failover in this situation?



More information about the pgpool-general mailing list