> We've been noticing some slowness with AFS3.4a and Solaris 2.5 (with
> all the recommended patches). Has anyone else seen this problem?
> Things seem to work fine for a while, then accessing certain files in AFS
> that aren't in the cache seems to take forever. For example, if I pick
> a path that I know probably isn't in the cache and truss an ls of it,
> ls will hang at the lstat call for a while (sometimes a couple of
> minutes), and then finally come back. This has happened on Sparc20s
> and UltraSPARCs.
I noticed something similar when I first started playing with a Sparc 5
running 2.5 as well. This was with stock Solaris 2.5, and the latest
AFS at the time, which I believe was the first incarnation of 3.4a.patches.
We use the Transarc-provided binaries for clients whenever possible.
When I did an ls of something in AFS, it would hang for rather longer than
I had seen on other systems. Now, this was primarily using -l or -F, so
it was statting every file. The strange part is that the slowness would
not go away once the directory in question was cached - an immediate ls
on the same directory would be just as slow.
Unfortunately, I can't reproduce the problem now - I just checked, and
it seems to have gone away. I haven't changed anything noticable - I'm
still using the same versions of both Solaris and AFS, though I've
rebuilt the machine a couple of times since them (always the same way,
though - I've been developing and testing our standardized build process
for sun4_55.
Somehow, I doubt this will be of much use to you, but I thought I'd
add a data point.
-- Jeffrey T. Hutzelman (N3NHS) <[EMAIL PROTECTED]>
Systems Programmer, CMU SCS Research Facility
Please send requests and problem reports to [EMAIL PROTECTED]
