Re: [squid-users] assertion failed

From what I can tell, those values whould be way more than needed. See
the "DISKD" chapter in the FAQ
<http://www.squid-cache.org/Doc/FAQ/FAQ-22.html>.

However, I am not a diskd user myself and cannot claim I am very familar
with exactly what it requires..

Hmm... reading some code.. looks like actually the internal diskd I/O
queue overflowed. Smells like a bug in diskd. It seems diskd does not
gracefully handle overload situations. This is what the Q1/A2 parameters
are meant to deal with, but apparently they are not foolproof..

What is your Q1 and Q2 values?

Regards
Henrik Nordström
Squid Hacker

Didi Rieder wrote:
>
> --On Thursday, November 01, 2001 08:28:00 PM +0100 Henrik Nordstrom
> <hno@squid-cache.org> wrote:
>
> > Your Squid ran out of shared memory for the communication to diskd.
> >
> > Increase the amount of shared memory available to Squid, or consider
> > switching to aufs/async-io instead of diskd.
>
> Thanks for your answer!
>
> Is there any way to estimate how much shared memory I should make available
> to Squid?
>
> My Sun machine has 2 gigs of mem, and 36 gig of cache space on disk. There
> are a 1000 users who use the proxy.
> The actual shm settings in /etc/system are the following:
>
> set shmsys:shminfo_shmmax=268435456
> set shmsys:shminfo_shmmin=100
> set shmsys:shminfo_shmmni=200
> set shmsys:shminfo_shmseg=100
>
> set semsys:seminfo_semmni=250
> set semsys:seminfo_semmns=500
> set semsys:seminfo_semmsl=250
> set semsys:seminfo_semmnu=500
> set semsys:seminfo_semume=50
>
> set msgsys:msginfo_msgmax=4096
> set msgsys:msginfo_msgmnb=16384
> set msgsys:msginfo_msgmni=100
> set msgsys:msginfo_msgtql=1024
>
> regards
> Didi
>
> --
> -------------------------
> Didi Rieder
> adrieder@sbox.tugraz.at
> -------------------------
Received on Thu Nov 01 2001 - 15:03:28 MST