[Pgcluster-general] Major problem, multiple thread update with placeholders hangs

[Tommi Berg]

Good day Hans-Juergen,

Hans-Juergen Schoenig wrote:
> this seems like a problem we had when we forked off Cybercluster and did
> some concurrency problem.
> if it is the same thing i think we fixed that. i am not sure the fix can
> be ported back to PGCluster however easily.
> if i remember correctly it might have been some locking issue.

That simple test case indeed works with Cybercluster. So I tried the
cybercluster too, but I got it to hang even at pgbench (10 scale factor,
10 threads). Debug message from pgreplicate at end of the mail. I also
got similar hang when trying my own perl script which ran updates at two
forks. And like the earlier mail from Lia Domine said, the loadbalancer
isn't using all three cluster databases, only first two. I'm sorry to
say, but I don't find Cybercluster 1.0.0 to be production ready either.


2008-01-31 11:25:33 [22607] DEBUG:replicate_packet_send_internal():
before PGRis_same_host_packet_tbl() from_host: clusterdb1 from_port: 5432
2008-01-31 11:25:33 [22607] DEBUG:replicate_packet_send_internal():
before PGRis_same_host_packet_tbl() actual host: clusterdb3 actual port:
5432
2008-01-31 11:25:33 [22607] DEBUG:getTransactionTbl(): thread_id=-1222853744
2008-01-31 11:25:33 [22607] DEBUG:start thread_send_cluster()
2008-01-31 11:25:33 [22607] DEBUG:getTransactionTbl(): thread_id=-1234433136
2008-01-31 11:25:33 [22607] DEBUG:getTransactionTbl(): thread_id=-1234433136
2008-01-31 11:25:33 [22607] DEBUG:thread_send_cluster(): end thread with
0 (CMD_TYPE_CONNECTION_CLOSE)
2008-01-31 11:25:33 [22607]
DEBUG:replicate_packet_send_internal():connect db:toba port:5432
user:toba host:clusterdb3 query:PGR_CLOSE_CONNECTION
2008-01-31 11:25:33 [22607] DEBUG:replicate_packet_send_internal():
thread_send_cluster thread created: -1239696496
2008-01-31 11:25:33 [22607] DEBUG:replicate_packet_send_internal():
waiting for thread -1234433136 to finish
2008-01-31 11:25:33 [22607] DEBUG:replicate_packet_send_internal():
waiting for thread -1239696496 to finish
2008-01-31 11:25:33 [22607] DEBUG:start thread_send_cluster()
2008-01-31 11:25:33 [22607] DEBUG:getTransactionTbl(): thread_id=-1239696496
2008-01-31 11:25:33 [22607] DEBUG:getTransactionTbl(): thread_id=-1239696496
2008-01-31 11:25:33 [22607] DEBUG:thread_send_cluster(): end thread with
0 (CMD_TYPE_CONNECTION_CLOSE)