So,
I am slowly upgrading the master slave infrastructure here.
I have two slaves, one is being prepped to become the new master, so it has no exclude tables, the other slave has some exclude tables.
both of these slaves are running centos 5.5 and mysql mysql Ver 14.12 Distrib 5.0.77,
the current master and the rest of the slaves are
Redhat Enterpriose ES 4 and mysql mysql Ver 14.12 Distrib 5.0.58,
So, on the big slave with no ignore-tables about every 20 minutes or once and hour
replication lag will spike from more or less no lag to 18040 seconds, just long enough to get a notification and then it goes away.
I was worried that it could be a disk i/o thing.. but iostat has the %util on the data volume pretty much pegged at 5% WITH THE REST OF THE UNDER 1% so it doesnt seem like a disk i/o thing. The load on the bos is under 1 pretty consistantly, even during or around the time it suddenly falls behind. The box is also not swapping.
I tried to blame the master, but the 4 other slaves are not having theis problem and it seems like if there was a master caused problem on one, it would be on all 4.
granted, there are tables the other slaves do not have....
So..
any thoughts on what I should be looking at next..
I need to get slow query logs on everything yet.
Ivan
I am slowly upgrading the master slave infrastructure here.
I have two slaves, one is being prepped to become the new master, so it has no exclude tables, the other slave has some exclude tables.
both of these slaves are running centos 5.5 and mysql mysql Ver 14.12 Distrib 5.0.77,
the current master and the rest of the slaves are
Redhat Enterpriose ES 4 and mysql mysql Ver 14.12 Distrib 5.0.58,
So, on the big slave with no ignore-tables about every 20 minutes or once and hour
replication lag will spike from more or less no lag to 18040 seconds, just long enough to get a notification and then it goes away.
I was worried that it could be a disk i/o thing.. but iostat has the %util on the data volume pretty much pegged at 5% WITH THE REST OF THE UNDER 1% so it doesnt seem like a disk i/o thing. The load on the bos is under 1 pretty consistantly, even during or around the time it suddenly falls behind. The box is also not swapping.
I tried to blame the master, but the 4 other slaves are not having theis problem and it seems like if there was a master caused problem on one, it would be on all 4.
granted, there are tables the other slaves do not have....
So..
any thoughts on what I should be looking at next..
I need to get slow query logs on everything yet.
Ivan