Hi Andrew

2.5.69-mm5 has some serious problems on my machine (pIII 700 704MB ram,
512MB swap). Every night when updatedb runs I get allocation failures, I
think it's mostly for received packets, they tend to get me disconnected
from all irc-servers etc. updatedb also tends to stall a lot, it can
take many hours for it to complete, I killed it after 8 hours once.

Swappiness is set to 60.

I've included procinfo and meminfo both from before I started updatedb
and when it stalled. I've also included vmstat output from two times it
stalled, they are quite diffrent, one has virtually no buffers and the
other one has lots, both have lots of free memory.

When it stalls vmstat says it's waiting on I/O, so does wchan but the
harddrive is very idle.

The allocation-problems I sometimes get during updatedb looks like this:

swapper: page allocation failure. order:2, mode:0x20
swapper: page allocation failure. order:2, mode:0x20
swapper: page allocation failure. order:2, mode:0x20
swapper: page allocation failure. order:2, mode:0x20
swapper: page allocation failure. order:2, mode:0x20


Here comes a lot of attachments, -fine is before updatedb was started,
-weird is the stall with lots of buffers (forgot to grab procinfo and
meminfo), -weird2 is the stall when there was virtually no buffers.

-- 
/Martin