Welcome back. You haven't been gone that long, only about 1.5 years.
Squid-2.4 is said to be on level with Squid-2.2.STABLE5, but
Squid-2.2.STABLE5-hno is probably still the most stable known Squid.
Async-I/O in 2.2.STABLE5-hno is probably also slightly faster than
Squid-2.4. In 2.2 writing of access.log and was also made async (if
async writes was configured), but this is not the case in Squid-2.4.
My current goal is to get Squid-2.5 fully stable, but unfortunately
there is other (non-Squid) tasks competing for my time at the moment...
Hopefully I can delegate most of the non-Squid tasks to my collegues,
allowing more time to be spent on Squid. I am more or less skipping
Squid versions 2.3 and 2.4.
Digest has and always will operate separately from ICP, with digests
having a higher priority than ICP.
Regarding statistics: Can you please elighten my mind on the issue..
exacly what statistics do you think is missing?
Regards
Henrik
David J N Begley wrote:
>
> Hi, Henrik; you probably wouldn't remember me - I was a regular of the
> squid-users mailing list for a few years some time ago. Hope you don't mind a
> quick question or two that require "expert" attention (thus I didn't just want
> uninformed opinion from the list as a whole).
>
> A number of sites, like us, have stuck with Squid 2.2.STABLE5 due to a number
> of reasons:
>
> - most primarily, we're running across multiple Fast/Wide SCSI, Ultra SCSI
> or Fibre Channel disks, under Solaris on SMP UltraSPARC boxes; we *rely*
> on a stable async I/O implementation and around the time of 2.3, you had
> mentioned that async I/O was being rewritten and thus had become somewhat
> less stable than 2.2
>
> - also, we exclusively use cache digests (no ICP) between sites across an
> ATM WAN and thus rely on stable peer digest requests, lookups and peer or
> parent up/down detection without the assistance of ICP; the only thing
> wrong with 2.2 regarding cache digests is that some statistics collection
> (for cachemgr) doesn't work (doesn't impair cache operation though)
>
> Could you give me some idea of where Squid is currently "at" as far as either
> 2.4-STABLE or 2.5-DEV is concerned? Is async I/O back to the performance and
> stability of 2.2 yet? Do cache digests still fully work without ICP? Has
> statistics collection all been fixed up?
>
> Thanks..
>
> ---
> "Training an African Grey parrot to function as an information systems
> manager can be very rewarding." - Philip Greenspun, MIT
Received on Mon Jun 11 2001 - 14:06:58 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:14:03 MST