On 11/27/15 5:45 PM, DrakoRod wrote:
> Hi everybody
>
> I've the next architecture with pgpool (streaming replication mode):
>
> 4 nodes
>
> 1 Master Node
> 2 Standbys Node
> 1 pgpool Node
>
> I've disabled the load balancing, because some clients report me problems
> with the load balancer, they told me the load balancer some times send
> querys to standby nodes, which has not yet recovered data and the querys
> fail, but this topic is for another thread.
>
> When I try run a stress test with hammerdb I see next errors in the pgpool
> Node
>
> 2015-11-27 16:48:21: pid 20190: FATAL: unable to read data from DB node 0
> 2015-11-27 16:48:21: pid 20190: DETAIL: EOF encountered with backend
> 2015-11-27 16:48:21: pid 19182: LOG: child process with pid: 20190 exits
> with status 256
> 2015-11-27 16:48:21: pid 19182: LOG: fork a new child process with pid:
> 20298
> 2015-11-27 16:48:21: pid 20163: FATAL: unable to read data from DB node 0
> 2015-11-27 16:48:21: pid 20163: DETAIL: EOF encountered with backend
> 2015-11-27 16:48:21: pid 19182: LOG: child process with pid: 20163 exits
> with status 256
> 2015-11-27 16:48:21: pid 19182: LOG: fork a new child process with pid:
> 20299
>
> In all Child, therefore the hammerdb stop the test because all connections
> it lots.
I've never run pgPool myself, but maybe this means the connection was
broken for some reason. Are there related errors in the Postgres server
log for that node?
--
Jim Nasby, Data Architect, Blue Treble Consulting, Austin TX
Experts in Analytics, Data Architecture and PostgreSQL
Data in Trouble? Get it in Treble! http://BlueTreble.com