>> 2) Since I keep running out of memory during big compiles I've
>> tried to add some swap space. The swapon man pages say,
>> "Swap over NFS may not work" and I'm beginning to agree
>> there. If I do the mkswap from the iPAQ I always get
>> "invalid swap file" from swapon. If I run mkswap from my
>> NFS server, then the swapon succeeds, but I get periodic
>> messages saying something like "rw_swap: invalid swap
>> file", (can't remember the exact message) and processes
>> start dying. Any hints from the skiffcluster maintainers
>> about how I can get this to work?
>
>We are using the loop back device, please look at /etc/inittab & /etc/fstab
>on the skiffcluster. This helps but we still have problems. Ken Causey has
>been making changes to /proc/sys/vm/freepages to make things work better,
>but I do not have the results.
I haven't done much more than play with the numbers a little bit on
skiffcluster5 which I've been using as a guinea pig. The results so
far would indicate that changing the defaults in /proc/sys/vm/freepages
can help, but so far I haven't hit an the magic numbers, if any exist.
Someone always seems to be able to come up with a package that won't
happily build (imlib for example). I'm not terribly knowledge on this
subject and I will be happy to listen to anyone who has any pointers.
One question for the list: We've been experiencing pauses in the NFS
system that can last several seconds; what effect if any would a several
second delay in a swap request have on the VM system?
Ken Causey
Received on Wed Oct 11 11:30:06 2000
This archive was generated by hypermail 2.1.8 : Tue May 04 2004 - 09:43:44 EDT