Jeremy,
We observed thousands of errors in the logs of one of our slave
servers suggesting that it was making repeated failed connections to
our master.
Example:
070905 20:15:21 [Note] Slave: received end packet from server,
apparent master shutdown:
070905 20:15:21 [Note] Slave I/O thread: Failed reading log event,
reconnecting to retry, log 'hlgbinlog-opera.000069' position 168356323
070905 20:15:21 [Note] Slave: connected to master
'[EMAIL PROTECTED]:3306',replication resumed in log
'hlgbinlog-opera.000069' at position 168356323
070905 20:15:21 [Note] Slave: received end packet from server,
apparent master shutdown:
070905 20:15:21 [Note] Slave I/O thread: Failed reading log event,
reconnecting to retry, log 'hlgbinlog-opera.000069' position 168356323
070905 20:15:21 [Note] Slave: connected to master
'[EMAIL PROTECTED]:3306',replication resumed in log
'hlgbinlog-opera.000069' at position 168356323
070905 20:15:21 [Note] Slave: received end packet from server,
apparent master shutdown:
070905 20:15:21 [Note] Slave I/O thread: Failed reading log event,
reconnecting to retry, log 'hlgbinlog-opera.000069' position 168356323
070905 20:15:21 [Note] Slave: connected to master
'[EMAIL PROTECTED]:3306',replication resumed in log
'hlgbinlog-opera.000069' at position 168356323
There were hundreds of these per second. A Google search found this:
http://bugs.mysql.com/bug.php?id=16927
This occurred shortly after the server ID conflict, and exactly when
the master was melting down.
David
On Sep 14, 2007, at 4:11 AM, Jeremy Cole wrote:
Hi David,
This sounds strange, and should NOT occur because of a server_id
conflict.
Regards,
Jeremy
David Schneider-Joseph wrote:
Thank you.
We had a situation recently where two slaves had a conflicting
server ID for several minutes, and shortly thereafter the master
started reporting errors which were indicative of data corruption
while executing queries. This happened as the CPU usage climbed
very rapidly, and ultimately the entire master machine crashed
with an out of memory error.
Does this sound like something that could have been caused by a
short- lived server ID conflict? All servers involved were
running 5.0.27.
Your answers would be most helpful!
Thanks,
David
On Sep 13, 2007, at 7:58 AM, Shawn Green wrote:
Hello David,
David Schneider-Joseph wrote:
Hi all,
What do you know about the effect of conflicting slave server
IDs on the master in general? And specifically, are you aware
of any issues with MySQL 5.0.27?
Your help is very much appreciated.
Thanks!
David
Repeating the same Server ID in your slave servers is "BAD". It
has caused minor problems like duplicate entries on the slaves
and major problems like over a TB of error logs in just a few
minutes (because of "failure to connect" errors). There are
several very good reasons why *each and every* server in a
replication setup needs its own, unique server_id. Many of them
are discussed in the chapter on Replication:
http://dev.mysql.com/doc/refman/5.0/en/replication.html
To see what has been fixed in MySQL since 5.0.27 was released,
please review the change logs documented here:
http://dev.mysql.com/doc/refman/5.0/en/releasenotes-cs-5-0.html
For a list of all other bugs (active and inactive) you are
invited to research the bugs database (it is a public forum) at:
http://bugs.mysql.com
--
Shawn Green, Support Engineer
MySQL Inc., USA, www.mysql.com
Office: Blountville, TN
__ ___ ___ ____ __
/ |/ /_ __/ __/ __ \/ /
/ /|_/ / // /\ \/ /_/ / /__
/_/ /_/\_, /___/\___\_\___/
<___/
Join the Quality Contribution Program Today!
http://dev.mysql.com/qualitycontribution.html
--
high performance mysql consulting
www.provenscaling.com
--
MySQL General Mailing List
For list archives: http://lists.mysql.com/mysql
To unsubscribe: http://lists.mysql.com/mysql?
[EMAIL PROTECTED]
--
MySQL General Mailing List
For list archives: http://lists.mysql.com/mysql
To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
