On 12/19/2005 7:56 PM, Marc G. Fournier wrote:
'k, setting up monitoring, and the script is reporting 1 out of 3 nodes
out of sync:
./check_slony_cluster.sh dns ams ams.hub.org
ERROR - 2 of 3 nodes not in sync
no problem, figured out in the script how it is being determined, and:
st_received | cfmdelay
-------------+-----------------
2 | 00:00:00.010721
3 | 03:59:55.2181
4 | 00:00:00.125318
(3 rows)
wow ... 3 hours and 59 minutes where the other two (Node 4 is a remote
server, somewhere in the US, while node 3 is the server beside the master)
...
Now, I've checked Node 3, and it contains the same # of records as Node 1
..
Now, I just did an update on one record in the table, and checked all 3
slaves and they see the change, yet now I'm seeing:
st_received | cfmdelay
-------------+-----------------
2 | 00:00:00.009916
3 | 03:59:55.175099
4 | 01:46:02.69134
(3 rows)
Node 4 just shot up ...
Looking at sl_status:
# select * from "_dns".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 | 837 | 2005-12-19 20:52:23.576685 |
837 | 2005-12-19 20:52:23.589583 | 2005-12-19 20:52:23.576685 |
0 | 00:00:06.669823
1 | 3 | 837 | 2005-12-19 20:52:23.576685 |
837 | 2005-12-20 00:52:18.736552 | 2005-12-19 20:52:23.576685 |
0 | 00:00:06.669823
1 | 4 | 837 | 2005-12-19 20:52:23.576685 |
837 | 2005-12-19 22:36:25.514229 | 2005-12-19 20:52:23.576685 |
0 | 00:00:06.669823
So, what is st_last_received_ts, and why isn't Node 3 updating it? I've
checked my slon_ams.out file on Node 3, and there are no errors being
generated that I can see ... and replication appears to be working fine on
all the Nodes ...
Somewhere else I need to be looking for this?
The timestamps st_last_event_ts (origin) and st_last_received_ts
(subscriber) are taken on different servers. If you play around with the
clocks of them you will find also settings where sl_status reports them
to be in the future.
Jan
--
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me. #
#================================================== [EMAIL PROTECTED] #
_______________________________________________
Slony1-general mailing list
[email protected]
http://gborg.postgresql.org/mailman/listinfo/slony1-general
