On Sun, 22 Jan 2012, Brian Fehrle wrote:

> Hi all,
>
> PostgreSQL 9.1.2
> Slony 2.1.0

Set max_stack_depth in your postgresql.conf to something higher.

sync_group_maxsize in your slon.conf to something low MIGHT help (ie 1 or 2) 
but I think the default in 2.1 is pretty low anyway (like 20).





>
> I am having some trouble getting a slon node caught up on events. It's a
> larger database, 350 or so Gigs, and I added a node to a replication set
> and while it was doing the initial sync, the server that the slon
> daemons were running on died. It wasn't until about 5 hours later we got
> the daemons running on a different node and it restarted (i assume it
> restarted) the initial sync.
>
> From what I can tell, it finished the initial sync, however now it's
> unable to catch up due to the following error line (reduced in size,
> don't know how many elements there actually were but the single line had
> about 18 million characters):
> 2012-01-22 04:43:07 EST ERROR  remoteWorkerThread_1: "declare LOG cursor
> for select log_origin, log_txid, log_tableid, log_actionseq,
> log_cmdtype, octet_length(log_cmddata), case when
> octet_length(log_cmddata) <= 1024 then log_cmddata else null end from
> "_myslonycluster".sl_log_1 where log_origin = 1 and log_tableid in
> (2,3,4,5,6,7,1,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122)
> and log_txid >= '34299501' and log_txid < '34311624' and
> "pg_catalog".txid_visible_in_snapshot(log_txid, '34311624:34311624:')
> and (  log_actionseq <> '2474682'  and  log_actionseq <> '2403310'  and
> log_actionseq <> '2427861'  and
> <SNIP, repeated many thousands of times with different numbers>
> '  and  log_actionseq <> '2520797'  and  log_actionseq <> '2519348'
> and  log_actionseq <> '2485828'  and  log_actionseq <> '2523367'  and
> log_actionseq <> '2469096'  and  log_actionseq <> '2520589'  and
> log_actionseq <> '2414071'  and  log_actionseq <> '2391417' ) order by
> log_actionseq" PGRES_FATAL_ERROR ERROR:  stack depth limit exceeded
>
> I found someone with a similar(ish) issue back in the day, and a
> function called compress_actionseq was mentioned. I turned up debugging
> to level 4 and see that it is indeed compressing the actionseq, and I
> looked at the code and it also looks like the above output IS the
> compressed sequence.
>
> Now, this seems to be a tricky setting to tweak on postgres, so I'd
> rather not unless I had to. So my thoughts were to hopefully just force
> slony to try to do smaller syncs at a time. I tried reducing (and for
> the heck of it increasing) the group size, desired_sync_time,
> sync_max_rowsize, and sync_max_largemem. However nothing has altered the
> size of this query that is being executed on the database.
>
> Any thoughts, suggestions? The initial sync of slony takes about 14
> hours, so I'd rather not drop the node and re-attach it. In fact I have
> two nodes in the same issue, stuck at the same event, so I'd rather just
> get them both synced up without doing another initial sync.
>
> Also, I toyed with the idea of forcing slon daemon to only sync up to a
> specific event, in hopes to do blocks of say 500 events, however the
> quit_sync_finalsync parameter is not accepted correctly by slony 2.1.0.
> (I've submitted a email to this list about this too).
>
> Thanks in advance,
> - Brian F
> _______________________________________________
> 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