It's probably a firewall timing out your PostgreSQL connection while the 
indexes are being built on the replica. 

Look into tcp keep alive settings. 

> On Feb 15, 2014, at 22:09, Tory M Blue <[email protected]> wrote:
> 
> 
> So I've been fighting with this for a few months. I had someone on slony Dev 
> attempt to lend a hand but others in the group, felt it was more of a  
> postgres issue. While this may be true, I'm still looking for some 
> assistance. Everything points to a disconnect in slony.
> 
> Wide area replication, fails on one of my largest tables. Now the table will 
> copy over complete no issues (using standard pgsql commands), it's the post 
> processing after the data is copied that seems to cause a sig term or 
> something on the connection, since slony states that the set failed and tries 
> again, fails at the same place ,
> 
> 2014-02-15 15:23:00 PST CONFIG remoteWorkerThread_1: Begin COPY of table 
> "tracking"."spotlightimp"
> 2014-02-15 16:46:45 PST CONFIG remoteWorkerThread_1: 5643041332 bytes copied 
> for table "tracking"."spotlightimp"   <--- Completes transfer
>  2014-02-15 17:34:10 PST CONFIG remoteWorkerThread_1: 7870.124 seconds to 
> copy table "tracking"."spotlightimp"    <-- At this point it finishes the 
> index creation and everything else
> 2014-02-15 17:34:10 PST CONFIG remoteWorkerThread_1: copy table 
> "tracking"."adimp"
> 2014-02-15 17:34:10 PST CONFIG remoteWorkerThread_1: Begin COPY of table 
> "tracking"."adimp"
> 2014-02-15 17:34:10 PST ERROR  remoteWorkerThread_1: "select 
> "_slonyschema".copyFields(19);"   <--- FAILS but adimp table is there, this 
> is a red herring. the issue is above!
> 2014-02-15 17:34:10 PST WARN   remoteWorkerThread_1: data copy for set 2 
> failed 1 times - sleep 15 seconds
> NOTICE:  Slony-I: Logswitch to sl_log_1 initiated
> CONTEXT:  SQL statement "SELECT "_slonyschema".logswitch_start()"
> PL/pgSQL function _slonyschema.cleanupevent(interval) line 96 at PERFORM
> 2014-02-15 17:34:14 PST INFO   cleanupThread: 7209.360 seconds for 
> cleanupEvent()
> 
> 
> I've brought my work_mem to over 40GB and that's not helping the length of 
> time for this large table. I have even removed the index statement still 
> doesn't cut the time,  The copy is fine, all the data comes over. It's 
> something in the processing of the table. There is s disconnect at some point 
> between when slony finishes up the copy of the spotlightimp, and Postgres 
> processes the rules in the table, and slony starts on the next table.
> 
> 
> 2014-02-15 18:48:27 PST CONFIG remoteWorkerThread_1: copy table 
> "tracking"."spotlightimp"
> 2014-02-15 18:48:27 PST CONFIG remoteWorkerThread_1: Begin COPY of table 
> "tracking"."spotlightimp"
> 2014-02-15 20:11:07 PST CONFIG remoteWorkerThread_1: 5643067207 bytes copied 
> for table "tracking"."spotlightimp"
> 2014-02-15 20:59:46 PST CONFIG remoteWorkerThread_1: 7878.124 seconds to copy 
> table "tracking"."spotlightimp"
> 2014-02-15 20:59:46 PST CONFIG remoteWorkerThread_1: copy table 
> "tracking"."adimp"
> 2014-02-15 20:59:46 PST CONFIG remoteWorkerThread_1: Begin COPY of table 
> "tracking"."adimp"
> 2014-02-15 20:59:46 PST ERROR  remoteWorkerThread_1: "select 
> "_slonyschema".copyFields(19);" 
> 2014-02-15 20:59:46 PST WARN   remoteWorkerThread_1: data copy for set 2 
> failed 1 times - sleep 15 seconds
> NOTICE:  Slony-I: log switch to sl_log_2 complete - truncate sl_log_1
> CONTEXT:  PL/pgSQL function _slonyschema.cleanupevent(interval) line 94 at 
> assignment
> 2014-02-15 20:59:50 PST INFO   cleanupThread: 7203.435 seconds for 
> cleanupEvent()
> 
> I do feel incredibly strongly it's the size of the table and how long the 
> process takes, the network / postgres is either reaping the connection or 
> other causing slony to be in an unknown state and causes the error the minute 
> we try to move forward from the spotlightimp table.. If I could cut down the 
> preprocessing after the table was copied that may solve it, but removing the 
> index part has not helped the situation as I hoped it would.  This is a 
> complicated table, as well as it's size.
> 
> I would love to get this sorted out, slony should allow for this remote 
> replication, but something is going wrong and man would I love to get this 
> resolved!
> 
> CentOS6.2
> Postgres 9.2.4 slony 2.1.3
> 
> Thanks
> Tory
> _______________________________________________
> Slony1-general mailing list
> [email protected]
> http://lists.slony.info/mailman/listinfo/slony1-general
_______________________________________________
Slony1-general mailing list
[email protected]
http://lists.slony.info/mailman/listinfo/slony1-general

Reply via email to