Vassiliki Iliadou wrote:
>
> Hi,
> I run squid 2.0patch1 on an HP (hp-ux 10.2).
> When I ask from cachemgr "General Runtime Information", squid stops to
> serve requests an I see the message below in the cache.log for all
> dnsservers :
>
> 1999/05/10 17:21:38| ipcCreate: fork: (12) Not enough space
> 1999/05/10 17:21:38| dnsOpenServers: WARNING: Failed to start
> 'dnsserver' #2.
> 1999/05/10 17:21:38| ipcCreate: fork: (12) Not enough space
> 1999/05/10 17:21:38| dnsOpenServers: WARNING: Failed to start
> 'dnsserver' #3.
> 1999/05/10 17:21:38| ipcCreate: fork: (12) Not enough space
> 1999/05/10 17:21:38| dnsOpenServers: WARNING: Failed to start
> 'dnsserver' #4.........
>
> This happens also when I reconfigure squids, so the squid stops
> running. Then I have to kill squid proccess and start it again so the
> dnsservers begin to run successfully.
>
> I noticed that all these above happen when the traffic is quite big.
> Can anybody help me please? Which space is not enough? Is it about swap
> space?
I'm guessing that squid is using more than 50% of your combined
ram+swap, and that HP-UX is not willing to overcommit memory during a
fork(). (fork() replicates the program, momentarily replicating it in
memory, and thus doubling it's memory requirements. IIRC, most systems
do not 'overcommit' memory, that is even if they save time by not
explicitly replicating pages immediately [linux fork()/vfork() as an
example] the pages are still allocated so that even if they were all to
suddenly be needed, the system would have enough to make good on the
promise it made when it said it was able to perform the requested
function)
Am I making any sense?
D
Received on Wed May 12 1999 - 15:21:09 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:46:16 MST