I have my data feeding to a Solaris Intel
machine with logging with no RAID..just three data disks
and the ldm.pq (thanks to suggestion by Art)
on the root disk. I have logging enabled
on all disks and it easily keeps up
with the full NOAAPORT feed (our SDI
here at Universal).
I think ext2 is very, very fast, but it's kind of like
a highly-tuned Ferrari..pretty fragile and a problem
will take you a long time to recover from.
It's also possible (I haven't tried it on our new Linux
box yet) that ext3 exacts a more severe penalty than
Solaris UFS+logging.
-----Original Message-----
Sent: Monday, July 22, 2002 12:18 PM
Cc: ldm-users@xxxxxxxxxxxxxxxx
On Mon, 22 Jul 2002, David Wojtowicz wrote:
> >Also, a word to Redhat 7.3 users: in case you haven't seen it, updated
> >psmisc packages are available
(http://rhn.redhat.com/errata/rh73-errata.html).
> >There is a bug whereby if you run ldmadmin stop, it sometimes won't
> >killall (or even any) of the processes. This patch fixes that problem.
> >Otherwise, and even with that, the LDM has never been more stable on my
> >system. RH 7.3 really rocks, by far and away the best release yet, IMO.
>
> NB... do not make the filesystem containing your product queue or
> destination of pqact entries an ext3 filesystem. You will pay
> severely in performance.
Hmmm... that's interesting. How much of a penalty are we talking about?
I could (should) put my product queue on an ext2 in that case, but my data
needs to be on an ext3 by necessity as I can't wait hours for my multi-GB
arrays to fsck if the system should crash or the power fails.
Art.
Arthur A. Person
Research Assistant, System Administrator
Penn State Department of Meteorology
email: person@xxxxxxxxxxxxxxxxxx, phone: 814-863-1563
