2016-10-30 - Re: [GRASE-Hotspot] Re: Monitoring performance - troubleshooting

Header Data

From: Glyn <bo***1@gmail.com>
Message Hash: 6c2a8b48387b42b31eef3208e131f7011cf7c0cab96906e67dc1626cc755b857
Message ID: <01267689-dd80-45bd-bc9d-2abc62fb52b4@grasehotspot.org>
Reply To: <CAESLx0LND53w1qQhCWVFcGpFn1p9bmWPWH2ZgJVM-P7xBGH2Hg@mail.gmail.com>
UTC Datetime: 2016-10-30 06:49:50 UTC
Raw Date: Sun, 30 Oct 2016 06:49:50 -0700

Raw message 

HI Tim

Thanks for the update and offer of help.

You are welcome to connect using the vpn and assist user, that is all 
installed as default so tell me what you need. I can also offer a direct 
ssh connection via a dynamic dns and obfuscated port number if that is any 
use, just let me know wher to send the addresses & credentials.

The system is on UK time and peak times are duing the day - lunchtime and 
early evening. Currently there only a few active sessions, but since 
installing it on Friday morning over 200 freeaccess users have been created.

Today I rebuilt the old laptop with ubuntu 14.04 and a new clean install of 
Grase, the idea being to see if the problem persists with that in place 
i.e. environmental or AP issues, the problem will be putting it in as the 
location of the new internet router is not really suitable so I want to do 
some more with the pi for now.

On Sunday, 30 October 2016 07:32:13 UTC, timwhite88 wrote:
>
> Hi Glyn
>
> Is it possible that there is something about the Pi architecture (it is 
> ARM not X86) that causes the issues? Its hard to know where to even look 
> for the performance issues given that I don't have a Pi to work with, or 
> lots of devices. I also expect that your testing of high bandwidth 
> applications is probably the wrong test. Lots of users who are browsing == 
> lots of small connections, not high bandwidth. I'm wondering if we are 
> hitting some limit with the number of connections. A good way to test would 
> be firstly to take Coova Chilli out of the equation, just have the Pi do 
> normal NAT. That would show if it's the extra CPU/IO of Coova chilli that's 
> at fault. (Yes, this would require no captive portal for awhile, but that 
> would be a good test).
>
> If you are happy to give me remote access (grase-conf-openvpn + 
> grase-remote-assist-user) and let me know the peak times (and timezone) I 
> can try and do some more monitoring to try work out what's happening. 
> Probably even setup Zabbix to monitor different aspects of the system to 
> see any correlation between the slowdown and other monitored items.
>
> I've not got heaps of time, and I understand you also don't have heaps of 
> time to debug it, but if you're willing to give it another shot, lets see 
> if we can work out the issue.
>
> Regards
>
> Tim
>
> On Fri, Oct 28, 2016 at 10:44 PM, Glyn <bo***.@gmail.com <javascript:>
> > wrote:
>
>> Just thought I'd update on this, but the news is not good. I am begining 
>> to doubt my sanity.
>> Two days ago I built a completely new installation on a new system with a 
>> known good usb / eth adaptor and this time using a SATA HD via a USB 
>> adaptor as I was worried that the overhead of accessing the sd card for 
>> logging an journalling might be slowing things down. At home I was able to 
>> stress test it with as many devices as I had using high bandwidth 
>> applications and is stood up to all this ust fine.
>> Today I tooke it to site and set things up and although it seemed OK soon 
>> ground to a halt. I found that about 20 users were active and I could 
>> barely get to the admin page anymore and no-one seemed to be getting any 
>> decent performance. Looking at it now from home I can see there are over 20 
>> active sessions right now and about 70 free access users have been created 
>> since I put it live 3 hours ago! looking at the sessions many of them are 
>> moving hardly any data and this correlates with the reports and my 
>> experience of it slowing down.
>> I firmly believe the Pi set up is fine now and am looking at tweeking the 
>> AP settings but I know that had worked for years with the old set up. The 
>> BT hub has wifi enabled and I use that as a back door on site and to access 
>> the settings to create my port forwarding rules to the Pi - so this tells 
>> me it is noet wifi interference per se.
>> I would welcome ay suggestions about where I can look next - it is not 
>> practical to go back to the old laptop solution long term because of where 
>> the router is but I might try a rebuild of that and see what happens - 
>> otherwise I am going to walk away and tell them to either open up the wifi 
>> on the router or buy a system.....
>>
>> -- 
>> 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/.
>> To view this discussion on the web visit 
>> https://groups.google.com/a/grasehotspot.org/d/msgid/grase-hotspot/b95e9ef8-6f35-44d3-a975-22a6fe4a5480%40grasehotspot.org 
>> <https://groups.google.com/a/grasehotspot.org/d/msgid/grase-hotspot/b95e9ef8-6f35-44d3-a975-22a6fe4a5480%40grasehotspot.org?utm_medium=email&utm_source=footer>
>> .
>>
>
>

Thread