I apologize if this is off-topic, because I'm not sure if
/proc/sys/vm/bdflush has anything to do with ext3 performance or not.
However, I've been searching around for a while and can't find the
answer I
need. If somebody could shed some light, I'd really appreciate it.
In "Securing and Optimizing Linux: RedHat Edition -A Hands on Guide",
the
author gives these values to use for bdflush:
# echo 100 1200 128 512 15 5000 500 1884 2 > /proc/sys/vm/bdflush
I've read the descriptions of what these values mean and think I understand
them. However, in another article, a particular reader gave feedback that he
thought these numbers were too aggressive because they could leave a lot of
data sitting in buffers waiting to be flushed to disk, which could cause
data corruption during an unclean shutdown. My question is, if I'm running
ext3 in "mode=journal", will the journaling aspects of ext3 protect me
even
if I tweak the bdflush params like this? Or are these apples and oranges?
(and maybe I should just leave the default bdflush params alone...)
I have two types of systems I'm configuring, hopefully for heavier load:
Single CPU Apache webservers with a single SCSI drive and 512MB RAM, and
dual CPU MySQL servers with two SCSI drives and 2GB RAM. Both types are
running kernel 2.4.18.
Thanks,
--jeff
