19.04.2013 23:39, Ilya Kazakevich:
Try to use "tuning-primer.sh": this scripts reads your variables and prints
memory size you need for that.
I tried that. The results are inconspicious:
MEMORY USAGE
Max Memory Ever Allocated : 5.27 G
Configured Max Per-thread Buffers : 1.92 G
Configured Max Global Buffers : 5.15 G
Configured Max Memory Limit : 7.07 G
Physical Memory : 22.98 G
Max memory limit seem to be within acceptable norms
Although the logics behind the "tuning primer" script are rather
simple and I understand predicting the memory usage for MySQL is
much harder:
http://www.mysqlperformanceblog.com/2012/03/21/troubleshooting-mysql-memory-usage/
20.04.2013 00:26, Rick James:
> What's the STATUS value of Threads_running? If it really is
> "~60-100 connection threads", then there could be any of a few
> temp allocations for the queries. Some allocations are
> per-subquery.
Usually around 2-4. I also tried checking if killing / resetting
existing (idle) connections would significantly reduce memory
usage when mysqld has reached ~20 GB - it would not, so this is
either not related to connection states or the memory is leaking
from there in a way which would be unaffected by closing the
connection.
> Is the system I/O bound? Or CPU bound? Or neither?
Neither - the system has plenty of headroom for both. The data
working set easily fits into the RAM, the amount of UPDATEs is
negligible (resulting in < 100 write requests per second for the
I/O subsystem). 1-minute load average is 2-3 under normal
(non-swapping) conditions with 6 CPU cores available.
> I recommend you optimize the queries.
I cannot do much about it. I am the infrastructure guy who is
"fixing the obviously broken DBMS". What I still cannot figure
out is if the behavior is due to a misconfiguration or a
regression / bug to file. And MySQL counters are not exactly
helping - it is completely opaque to me where the memory is going.
--
Denis Jedig
--
MySQL General Mailing List
For list archives: http://lists.mysql.com/mysql
To unsubscribe: http://lists.mysql.com/mysql