Hi All,
We are having a very strange problem with two of our slave servers,
here is what is happening
mysql> select version();
5.0.30-Debian_1~bpo.1-log
1) we backup our slave every night, and when we do that we stop the slave,
and start a remote mysqldump from the backup machine..
2) within usually the first 5 minutes we get the one or more of the
following errors in the error log
070701 11:19:50 [ERROR] Error reading packet from server: Lost connection to
MySQL server during query ( server_errno=2013)
070701 11:19:50 [Note] Slave I/O thread: Failed reading log event,
reconnecting to retry, log 'mysql-bin.022362' position 1024154
070701 11:19:50 [Note] Slave: connected to master '
[EMAIL PROTECTED]:3306',replication resumed in log '
mysql-bin.022362' at position 1024154
3) and after the backup is done, when the sql_thread is started back, it
results in a "SQL Syntax error", thereby stopping replication...
4) and each time this happens the very last statement in the relay log
before mysql rotates the log file after reconnecting to the master (see
above error) , usually has garbage characters or some sql statements
completely messed up like the one below
INSERT INTO clientsession_events_data
(clientsession_events_id, data)
VALUES
(xxxxxxx, 'array (\n
\'memory_stats\' => \n array (\n \'native_thread_count\' => 2 VALUES
(xxxxxx, 9, NOW(),
0, '')
/* /catalog/skudb/test3.php */[EMAIL PROTECTED]
Èê/@[EMAIL PROTECTED]"@SYSTEMcustomer_xxxx/*shard
customer://14347061/reporting */
INSERT INTO clientsession_eve;
# at 1024291
#070701 11:19:51 server id 1294 end_log_pos 1024343 Rotate to
AF001294-relay-bin.007930 pos: 4
5) I can easily reproduce this error and as I was telling usually during the
first few minutes of starting a backup or when the slave IO_Thread is behind
and is reading a lot of Bin logs from the master ....
I have seen this before in one of the posts, but there didn't seem to be a
solution , I guess there was also a big that was filed ...
I ran some ping tests to/from this machine and there was no packet loss, and
strange thing is if network was a problem it should be happening all the
time....
mysql> show global variables like "%timeout%";
+--------------------------+-------+
| Variable_name | Value |
+--------------------------+-------+
| connect_timeout | 5 |
| delayed_insert_timeout | 300 |
| innodb_lock_wait_timeout | 50 |
| interactive_timeout | 120 |
| net_read_timeout | 5 |
| net_write_timeout | 5 |
| slave_net_timeout | 3600 |
| table_lock_wait_timeout | 50 |
| wait_timeout | 3 |
+--------------------------+-------+
Anybody has any ideas, or have encountered this before .....
-Kishore Jalleda
http://kjalleda.googlepages.com
