I'm running mysql 4.0.25 on netbsd 3, on a dual-processor opteron machine with
16GB. I'm trying to make things faster, of course. All the data operations are
on one very large table (about 20GB, index is 17GB), which I will eventually
figure out how to split.
I have four processes working on this table at once; they select 20,000 or more
row, chew for a while, and then update those rows. I have a lot of trouble with
the index blocks getting flushed when updates are done in parallel, so there is
some judicious locking to help that. (The key_buffer_size is set to 4GB, which
is the maximum supported - I'd use more if I could.)
What I notice is that even though I have two client threads doing SELECT at the
same time, I never manage to use more than one CPU, maybe dipping into the
second one a small bit. The mysql config log indicates that it found pthreads
and compiled with it. systat vmstat tells me that the disks are idle, memory is
full, and top indicates that the mysqld process is CPU bound. There are many
many syscalls being completed; I'm guessing that the blocks are moving from file
buffer pages to the process.
Is there a tool (or option to ps or top) that lets me view the thread activity
inside a process?
I see in the pthread(3) man page that there's an environment variable
PTHREAD_CONCURRENCY The number of concurrent threads to be run.
This value should at least be 1, and smaller
than or equal to the number of CPUs.
Do I need to set this to be more than 1 in the mysqld environment?
Thanks,
chris
--
MySQL General Mailing List
For list archives: http://lists.mysql.com/mysql
To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
