2015-02-03 23:16:20 |
mig5 |
description |
I have a server running Debian 6 (Squeeze) and Percona 5.5.
Logs indicate that for a while, there's been a warning that:
150202 22:19:18 InnoDB: ERROR: the age of the last checkpoint is 9448357,
InnoDB: which exceeds the log group capacity 9433498.
InnoDB: If you are using big BLOB or TEXT rows, you must set the
InnoDB: combined size of log files at least 10 times bigger than the
InnoDB: largest such row.
I am aware of what this means and that the innodb_log_buffer_size needs to be increased. Unfortunately no-one had noticed this alert on this server :/
But since the upgrade to 5.5.41-rel37.0-727.squeeze, eventually these error alerts in the log seem to culminate in a giant MySQL crash, which didn't used to happen. These lines, for example, appear right after the last checkpoint alert as above (but there had been many preceding it):
150202 22:19:18 InnoDB: Assertion failure in thread 1420909424 in file sync0rw.c line 569
InnoDB: Failing assertion: !lock->recursive
InnoDB: We intentionally generate a memory trap.
A full output of the log is attached (from the last checkpoint error, and onward). The corruption appeared to fix itself on this occasion (without anyone even noticing), but the next day it occurred again and we ended up with a totally corrupt database system after this. All I know is that I've not seen MySQL fatally crash on something like the checkpoint alert before.
System is 3.18.3-x86_64-linode51 #1 SMP Fri Jan 23 09:57:22 EST 2015 x86_64 GNU/Linux |
I have a server running Debian 6 (Squeeze) and Percona 5.5.
Logs indicate that for a while, there's been a warning that:
150202 22:19:18 InnoDB: ERROR: the age of the last checkpoint is 9448357,
InnoDB: which exceeds the log group capacity 9433498.
InnoDB: If you are using big BLOB or TEXT rows, you must set the
InnoDB: combined size of log files at least 10 times bigger than the
InnoDB: largest such row.
I am aware of what this means and that the innodb_log_buffer_size needs to be increased. Unfortunately no-one had noticed this alert on this server :/
But since the upgrade to 5.5.41-rel37.0-727.squeeze, eventually these error alerts in the log seem to culminate in a giant MySQL crash, which didn't used to happen. These lines, for example, appear right after the last checkpoint alert as above (but there had been many preceding it):
150202 22:19:18 InnoDB: Assertion failure in thread 1420909424 in file sync0rw.c line 569
InnoDB: Failing assertion: !lock->recursive
InnoDB: We intentionally generate a memory trap.
A full output of the log is attached (from the last checkpoint error, and onward). The corruption appeared to fix itself on this occasion (without anyone even noticing), but the next day it occurred again and we ended up with a totally corrupt database system after this. All I know is that I've not seen MySQL fatally crash on something like the checkpoint alert before. So unsure if it's a regression in the latest Percona 5.5, which we'd upgrade to a few days before.
System is 3.18.3-x86_64-linode51 #1 SMP Fri Jan 23 09:57:22 EST 2015 x86_64 GNU/Linux |
|