On May 29, 2008, at 12:36 PM, Jan Wieck wrote:
Over the last 30 hours there have been a tremendous amount of
changes against my db. This morning there are 5.6 million entries
in sl_log_1. Over time the fetch log statements are getting
longer in duration.
Is there any hope the log switch/truncate will complete?
A log switch can be held up indefinitely by one single node not
catching up or at least its event confirmations not making it back
to the origin. Do you see anything suspicious in sl_status on the
origin?
Jan
I don't see anything suspicious. Other that my lagtime has almost
doubled.
mycluster=# SELECT * from _mycluster.sl_status ;
st_origin | st_received | st_last_event | st_last_event_ts
| st_last_received | st_last_received_ts |
st_last_received_event_ts | st_lag_num_events | st_lag_time
-----------+-------------+---------------+----------------------------
+------------------+----------------------------
+----------------------------+-------------------+-----------------
1 | 2 | 659998 | 2008-05-29 13:11:49.256412
| 659996 | 2008-05-29 13:11:27.181933 | 2008-05-29
13:11:33.224593 | 2 | 00:00:19.023239
(1 row)
mycluster=# SELECT * from _mycluster.sl_event_seq ;
sequence_name | last_value | increment_by | max_value |
min_value | cache_value | log_cnt | is_cycled | is_called
---------------+------------+--------------+---------------------
+-----------+-------------+---------+-----------+-----------
sl_event_seq | 659998 | 1 | 9223372036854775807
| 1 | 1 | 12 | f
Thanks,
bruce
LOG: 00000: duration: 7768.196 ms statement: fetch 100 from LOG;
LOCATION: exec_simple_query, postgres.c:1015
LOG: 00000: duration: 7775.775 ms statement: fetch 100 from LOG;
LOCATION: exec_simple_query, postgres.c:1015
LOG: 00000: duration: 7764.928 ms statement: fetch 100 from LOG;
LOCATION: exec_simple_query, postgres.c:1015
STATEMENT: delete from "_mycluster_procluster".sl_log_1 where
log_origin = '1' and log_xid < '252358837'; delete from
"_mycluster_procluster".sl_log_2 where log_origin = '1' and log_xid
< '252358837'; delete from "_mycluster_procluster".sl_seqlog
where seql_origin = '1' and seql_ev_seqno < '647367'; select
"_mycluster_procluster".logswitch_finish();
NOTICE: 00000: Slony-I: log switch to sl_log_2 still in progress
- sl_log_1 not truncated
LOCATION: exec_stmt_raise, pl_exec.c:2149
STATEMENT: delete from "_mycluster_procluster".sl_log_1 where
log_origin = '2' and log_xid < '220973324'; delete from
"_mycluster_procluster".sl_log_2 where log_origin = '2' and log_xid
< '220973324'; delete from "_mycluster_procluster".sl_seqlog
where seql_origin = '2' and seql_ev_seqno < '605163'; select
"_mycluster_procluster".logswitch_finish();
Thanks in advance,
bruce
_______________________________________________
Slony1-general mailing list
[email protected]
http://lists.slony.info/mailman/listinfo/slony1-general
