2016-11-23 - little help for an issue on a old 3.7.7.6 version

Header Data

From: Gianluca Filippini <gi***4@gmail.com>
Message Hash: f6d9607187519bc447e25e55cb89fe78576565e737e1ec745777c19ed4687820
Message ID: <6a21adae-f758-452b-9a0e-3a7544a089dd@grasehotspot.org>
Reply To: N/A
UTC Datetime: 2016-11-23 10:29:53 UTC
Raw Date: Wed, 23 Nov 2016 09:29:53 -0800

Raw message 

Hi Tim (and list),
I have an issue on a 3.7.7.6 version on a (old) machine that is serving 
~120 users in a church .. and I was looking for some guidance (I know a LOT 
has changed since then and I do plan to upgrade to 3.8 soon, but I wanted 
to understand the problem before upgrading in "blind" mode)

the grase machine is a 4core 3Ghz with 16G of ram, up from 2013, behind a 
firewall (IPFire).
both the grase and fw machines have the same configuration, but while the 
fw is updated frequently the grase machine is still running 3.7.7.6 on a 
ubuntu 11.10

this has been stable for a long, long time, but recentlyy I noticed that 
somehow the users are getting poor performance in terms of browsing:
the user does the login, request a page ... and it takes 10/15seconds to 
get the page you need.

I started to check  the wifi AP and they are fine, I checked the fw (up to 
date) and it is fine.

every time this is happening I noticed that I simply reboot the grase 
machine and everything is fast again, 
no delay in service.

I monitored CPU, ram and hdisk for a while and the grase/ubuntu box is 
basically empty, no significant traffic load to stress the hw.

===
so here I am looking for some hint on how to properly debug this 
intermittent problem...

I doubted  Squid3 was the culprit (but I know by default it should not 
cache at all) and I wanted to disable the proxy completely.

I followed the steps of this post
https://groups.google.com/a/grasehotspot.org/forum/#!topic/grase-hotspot/gl2uRgHmw9c

but somehow if I turn off squid, change the ipup.sh chilli script and 
restart the chilli service no one from the hotspot can surf the web anymore
:-(


===
any idea on what should I double check when this is happening?

I already check all the APs (6 access point on a gigabit) and the latency 
for each device is <1ms,
did an iperf test and each AP is verified to support gigabit on eth (these 
are ubiquiti AP dual band)


thanks in advance.

Thread