RE: [squid-users] cache_swap_low and high waternmark

From: Jigar Rasalawala <jrasalawala@dont-contact.us>
Date: Thu, 1 May 2003 18:38:22 -0700

Hi, Henrik

I am using Squid-2.5.STABL2 on RedHat 7.3.
I found another interesting things here.

Two snmp object that Squid supports seam to reach a numeric value and then
start going backwards. The two Objects are:

   .1.3.6.1.4.1.3495.1.3.1.7.0 = Counter32:
   .1.3.6.1.4.1.3495.1.3.1.9.0 = Counter32:

   Here is some output from a script I wrote that polls the first Object every
15 seconds:

.1.3.6.1.4.1.3495.1.3.1.7.0 = Counter32: 9252
.1.3.6.1.4.1.3495.1.3.1.7.0 = Counter32: 9281
.1.3.6.1.4.1.3495.1.3.1.7.0 = Counter32: 9301
.1.3.6.1.4.1.3495.1.3.1.7.0 = Counter32: 9327
.1.3.6.1.4.1.3495.1.3.1.7.0 = Counter32: 9350
.1.3.6.1.4.1.3495.1.3.1.7.0 = Counter32: 9374
.1.3.6.1.4.1.3495.1.3.1.7.0 = Counter32: 9395
.1.3.6.1.4.1.3495.1.3.1.7.0 = Counter32: 9412 <<==== Here starts going down?
.1.3.6.1.4.1.3495.1.3.1.7.0 = Counter32: 9372
.1.3.6.1.4.1.3495.1.3.1.7.0 = Counter32: 9322
.1.3.6.1.4.1.3495.1.3.1.7.0 = Counter32: 9262
.1.3.6.1.4.1.3495.1.3.1.7.0 = Counter32: 9170
.1.3.6.1.4.1.3495.1.3.1.7.0 = Counter32: 9097
.1.3.6.1.4.1.3495.1.3.1.7.0 = Counter32: 8974

   This may just be an SNMP issue or it could be an internal Squid problem.
If SNMP is just getting the values from Squid internals, then this could explain
many things. If this is a rollover then is should not happen until the
Counter23 hits 4294967295.

   A point of interest, this happens right at the spot that Squid hits his
low_water_mark. In this case, 90%.

If you have any clue what is going on at that time ?

Thank you
Jigar

-----Original Message-----
From: Henrik Nordstrom [mailto:hno@squid-cache.org]
Sent: Thursday, May 01, 2003 3:40 PM
To: Jigar Rasalawala; squid-users@squid-cache.org
Subject: Re: [squid-users] cache_swap_low and high waternmark

On Thursday 01 May 2003 23.55, Jigar Rasalawala wrote:

> If I want to support 50-70 Mbps, how much RAM and disk space should
> I use ? roughly. Is there any documentation available on that ? Can
> u suggest me a link ?

Going very much above 30 Mbps of HTTP traffic with a single Squid
server is not very practical. For such configurations it is
recommended to use a farm of several servers, each with 4 drives for
cache. For 50-70 Mbps I would use a cluster of 3-4 servers, where one
of the servers optionally also acts as load balancer with LVS if you
do not have a separate load balancer available (optionally HA-enabled
by LVS clustering with one of the other cache servers).

Note: You will probably not be able to use all space on the cache
drives. The number of drives is not for space but for the number of
heads to reduce seek time. See the Squid FAQ on memory usage for
cache space/memory dimensioning.

If this traffic is not only HTTP traffic but lots of other traffic
such as P2P file sharing, email, native FTP etc then the calcualtions
will obviously be different. Squid is only involved on HTTP or HTTP
proxied traffic (i.e. ftp by HTTP clients configured to use Squid as
proxy), not other unrelated traffic.

Regards
Henrik

-- 
Donations welcome if you consider my Free Squid support helpful.
https://www.paypal.com/xclick/business=hno%40squid-cache.org
If you need commercial Squid support or cost effective Squid or
firewall appliances please refer to MARA Systems AB, Sweden
http://www.marasystems.com/, info@marasystems.com
Received on Thu May 01 2003 - 19:39:05 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:16:09 MST