2016-11-30 - Re: [GRASE-Hotspot] question on GRASE freeradius temp file

Header Data

From: Dražen Žuvela <dr***a@radez.hr>
Message Hash: 64a3d6d076b0910f729e0b4c290d0fde805e5677b162645e8ce7806b187f531e
Message ID: <7136dca7-d9a9-9ed5-0ad4-91e2a568fd1a@radez.hr>
Reply To: <cbf014e4-5794-4616-81a0-80e46dead5bd@grasehotspot.org>
UTC Datetime: 2016-11-30 01:50:44 UTC
Raw Date: Wed, 30 Nov 2016 09:50:44 +0100

Raw message 

Hi Gianluca
Just to be sure disk space is OK try "df" at console. Look for any 
partition which may be full or near full.

Rgds.
Drazen

29.11.2016. u 23:52, Gianluca Filippini je napisao/la:
> thanks a lot for your quick reply Tim!
>
> mmm .. weird .. If I just remove that line in the default.grase config 
> and reboot the machine freeradius does NOT come up anymore.
> I put back that line .. and it works again... really strange
> (I know .. I can just restart the service, just want to make sure 
> everything works on a cold start)
>
> regarding the user load: this "delay" issue has been a complain from a 
> long time, but I cannot pinpoint if there was a jump.
>
> definitely I saw the problem myself ... user is stuck waiting for 
> minutes .. and then suddenly the page arrives really fast!
>
> the internet speed is tested to be 80/90Mbit/s ... the only bottleneck 
> is the grase linux box,
> so either something related to freeradius checking for concurrent 
> sessions or squid3 having troubles,
> but afaik we have cache disabled by default in squid3 .. so that 
> should not delay the query/response.
>
> the fact is that today I did a reboot when the issue was there ... and 
> suddenly surfing was fast again,
> so ... something related to some "incremental/overtime" issue
>
> at the time of the problem cpu load was 2%, memory was 20%, no swap
> :-(
>
>
>
>
>
> On Tuesday, November 29, 2016 at 9:53:52 PM UTC+1, timwhite88 wrote:
>
>     Hi Gianluca
>
>     We use the raddact table to detect multiple logins. Looking at
>     https://github.com/GraseHotspot/grase-conf-freeradius/blob/master/freeradius/sites-available/default.grase#L464
>     <https://github.com/GraseHotspot/grase-conf-freeradius/blob/master/freeradius/sites-available/default.grase#L464>
>     we can comment out that line which should prevent the radwtmp file
>     being written at all. You'll find that file in
>     /etc/freeradius/sites-available/
>     I'll aim to fix that in the package shortly.
>
>     Let us know if that fixes your issue. If it's recently become an
>     issue, I'd be looking at the volume of users you are getting, and
>     see if that has increased recently.
>
>     Regards
>
>     Tim
>
>     On Wed, Nov 30, 2016 at 5:24 AM, Gianluca Filippini
>     <gi***.@gmail.com <javascript:>> wrote:
>
>         Hi Tim,
>         in a previous post I was referring to poor performance on a
>         old installation of grase (3.7)
>
>         the issue is presenting as a long delay from the moment a
>         registered user asks for a website and the moment he gets the
>         page.
>         the delay is huge ... up to minutes ... on a very powerful
>         machine (4 cores 3Ghz Xeon with 16G ram) and a very fast
>         internet link (100Mbit/s fiber)
>         the average internet load is 90 users connected with a
>         20/30Mbit/s traffic.
>
>         so I focused my attention on the freeradius package and I
>         found that the var/log/freeradius/radwtmp was getting huge
>         after few days ... up to 600M+
>
>         my understanding is that this file is used by freeradius to
>         detect MULTIPLE SESSION which indeed I do *not* allow in my
>         GRASE user account group.
>
>         ===
>         My theory is that every time a user does a request this file
>         has to be parsed to prevent double sessions, adding a big
>         delay to the user response,
>         so I deleted the file (it is a temp file) and I will monitor
>         performance next .. but ..
>
>         is this some sort of known issue that has been addressed in
>         v.3.8 ?
>
>         I see that multiple session can be adressed either by using
>         mysql (raddact table) OR radwtmp file ..
>         do you know which one we are using?
>
>         thanks
>         -- 
>         This mailing list is for the Grase Hotspot Project
>         http://grasehotspot.org
>         ---
>         You received this message because you are subscribed to the
>         Google Groups "Grase Hotspot" group.
>         To unsubscribe from this group and stop receiving emails from
>         it, send an email to gr***.@grasehotspot.org
>         <javascript:>.
>         To post to this group, send email to
>         gr***.@grasehotspot.org <javascript:>.
>         Visit this group at
>         https://groups.google.com/a/grasehotspot.org/group/grase-hotspot/
>         <https://groups.google.com/a/grasehotspot.org/group/grase-hotspot/>.
>         To view this discussion on the web visit
>         https://groups.google.com/a/grasehotspot.org/d/msgid/grase-hotspot/3153bc50-6244-4d92-a9a9-0e00725444e2%40grasehotspot.org
>         <https://groups.google.com/a/grasehotspot.org/d/msgid/grase-hotspot/3153bc50-6244-4d92-a9a9-0e00725444e2%40grasehotspot.org?utm_medium=email&utm_source=footer>.
>
>
> -- 
> This mailing list is for the Grase Hotspot Project http://grasehotspot.org
> ---
> You received this message because you are subscribed to the Google 
> Groups "Grase Hotspot" group.
> To unsubscribe from this group and stop receiving emails from it, send 
> an email to gr***e@grasehotspot.org 
> <mailto:gr***e@grasehotspot.org>.
> To post to this group, send email to gr***t@grasehotspot.org 
> <mailto:gr***t@grasehotspot.org>.
> Visit this group at 
> https://groups.google.com/a/grasehotspot.org/group/grase-hotspot/.
> To view this discussion on the web visit 
> https://groups.google.com/a/grasehotspot.org/d/msgid/grase-hotspot/cbf014e4-5794-4616-81a0-80e46dead5bd%40grasehotspot.org 
> <https://groups.google.com/a/grasehotspot.org/d/msgid/grase-hotspot/cbf014e4-5794-4616-81a0-80e46dead5bd%40grasehotspot.org?utm_medium=email&utm_source=footer>.

Thread