Re: HUGE size of squid binary after compile

Yung-Zen Lai wrote:
>
> But my Squid seldom crashes. :p (I won't strip them from now on :) )
> The situation I met was out of space in my H.D
> : diskHandleWrite: FD 71: disk write error: (28) No space left on device
> : storeSwapOutHandle: SwapOut failure (err code = -6).
>
> BUT. when I use the command "df" to check my disk.
> I found that the capacity in my filesystem is only 86%, and i-node is 7%
> Would you please tell me how to find errors in it?

Your systems messages file may provide additional information on this.
See /var/adm/messages or /var/log/messages.

There are a couple of known causes to this:

* You are using Solaris, or another system where df reports free block
fragments and not free blocks.

Recommended action: Set filesystem optimisation to space and not time
(see tunefs manpage).

* You are using Solaris 2.6 and your mkfs has built a filesystem with
way to large cylinder groups..

Recommended action: Run newfs again, with a more approriate cylinder
group specifiaction (see Squid-users archives). WARNING: This wipes your
cache disk clean.

* You have a harware failure

Recommended action: Replace failing drive.

* You are bitten by a Squid bug that causes Squid to temporarily fill
the drive on certain requests.

Recommended action: Try to find which requests that fills your drive,
and complain to squid-bugs@ircache.net once you have found one such
request.

---
Henrik Nordstrom
Spare time Squid hacker