Fri Jan 9 14:46:19 PST 2009
- Previous message: [Slony1-general] Re: [Slony1-hackers] www.slony.info scheduled maintenance
- Next message: [Slony1-general] Re: Slony1-general Digest, Vol 23, Issue 7
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
All, I've been using slony 2.0 (RC2 and RC1...now on 2.0REL) for a while now replicating a single table. Things were going okay then it started to slow down and my slave db got out of sync, I tweaked some parameters on the master and slave to get the slave "caught" up with replication and things are okay for now. One thing I noticed in the master node is that I keep getting the 'log switch to sl_log_2 still in progress - sl_log_1 not truncated' message, and my sl_log_2 table is around 12GB (sl_log_1 is small). I'm getting around a 30-40 second delay to first row in the log files (on the slave), which was .001 seconds when I first started replicating several months ago. This growth in time makes sense since the table is getting larger and larger. From what I can tell the logs should flip back and forth from sl_log_1 and sl_log_2 and truncate the old logs to keep these tables to a minimum size and thus it shouldn't take 30 seconds to get all the SYNC events. My question is this: What are the steps needed to unwedge this log switch scenario, I can't see any obvious functions to call to force it. From what I can tell the two tables are in sync. sl_status() shows lag of 1-2 If I'm understanding the system correctly the logswitch code is waiting for the transactions that remain in the 'old' log to be taken care of before it truncates and complets the switch to log_2. Is there something that can tell the slon daemons to take a look at these dangling transactions, or a function to flush(delete) them? Or is starting completely over the solution? I've looked through the docs and other archives on how to troubleshoot this and didn't find anything to usefull. I've REINDEX'd the tables, vacuumed them just for good measure and I still see the same amount of time to perform the sync. There is also a chance that there could be things hanging around if there were issues with the other release candidates, this is the second time the issue happend. The first time it happened I just dropped the slave node and resynced to a new db. I'm willing to setup everything from scratch again. I just want to know what to do if I run into this situation again after starting over. Software setup: Ubuntu linux 7.10 Master: Postgres 8.3.3 Slave: Postgres 8.3.1 Slony: 2.0.0 Thanks in advance. -Steve
- Previous message: [Slony1-general] Re: [Slony1-hackers] www.slony.info scheduled maintenance
- Next message: [Slony1-general] Re: Slony1-general Digest, Vol 23, Issue 7
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Slony1-general mailing list