Re: delay pool not working right?

From: David Luyer <luyer@dont-contact.us>
Date: Thu, 01 Jul 1999 13:05:23 +0800

> I have no sibling hosts configured, and DNS traffic seems so small.
> So, there may be many overheads on HTTP.
> I will try to measure contents of traffic there in a few days.
>
> I tried to reduce these figures to slow things more tightly, but it
> causes frequent timeouts on client browsers. I wonder if it causes
> problem that I have only one child of the squid proxy, on which I
> configured delay_pool. Does the aggregate buckets of delay_pool work
> fine on this condition?

I think you might be running out of file descriptors, but I can't
be sure from just that info.

The aggregate should always work, but you have to be aware of the issue
mentioned a few times on the mailing list - it is possible for some
connections to get starved while others are served due to the way in
which squid scans file descriptors. I have finally decided what I think
should be done here, now it's just a question of implementing it.
(well, I think I have decided anyway, to keep a last read time on each
file descriptor [if there isn't one already] or to keep the file descriptors
on a linked list ordered by last read, and to walk through them in that order
rather than in plain numerical order... actually I'm not completely sure
on this either :-()

> Another question...
> My cachemgr.cgi shows
>
> Current: -18
>
> What does it mean?

This represents the amount of traffic currently in the 'pool'.

If the value is less than or equal to 0, it means request(s) are currently
being delayed.

David.
Received on Wed Jun 30 1999 - 22:47:35 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:47:04 MST