Hospot having issues?

[cdavis]

v1.13.3 build 3993

I had a call from a client who could not get online, she was not getting an IP. I switched and watched the radius log and it was looping? Here is the log (see below for more info)I

Fri Feb 13 14:31:10 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 14:31:11 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 14:31:12 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 14:31:12 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 14:31:16 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 14:31:16 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 14:31:17 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 14:31:18 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 14:31:19 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 14:31:19 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 14:31:20 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 14:31:20 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 14:31:21 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 14:31:22 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 14:31:23 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 14:31:23 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 14:31:23 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 14:31:24 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 14:31:25 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 14:31:26 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 14:31:26 2004 : Auth: Login OK: [00:08:0d:db:d3:60/00:08:0d:db:d3:60] (from client castle port 0 cli 0.0.0.0)


I noticed that the processor was 100%, so I rebooted. After reboot her machine got an IP and all worked perfectly. Processor was OK. I still saw two Login OK: lines for her though. Then about 30 minutes later I had a different person call. Same problem, I looked, processor was 100% again. Here is the log from that user:

Fri Feb 13 17:06:30 2004 : Auth: Login OK: [00:c0:a8:86:93:3f/00:c0:a8:86:93:3f] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 17:06:31 2004 : Auth: Login OK: [00:c0:a8:86:93:3f/00:c0:a8:86:93:3f] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 17:06:31 2004 : Auth: Login OK: [00:c0:a8:86:93:3f/00:c0:a8:86:93:3f] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 17:06:33 2004 : Auth: Login OK: [00:c0:a8:86:93:3f/00:c0:a8:86:93:3f] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 17:06:33 2004 : Auth: Login OK: [00:c0:a8:86:93:3f/00:c0:a8:86:93:3f] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 17:06:34 2004 : Auth: Login OK: [00:c0:a8:86:93:3f/00:c0:a8:86:93:3f] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 17:06:34 2004 : Auth: Login OK: [00:c0:a8:86:93:3f/00:c0:a8:86:93:3f] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 17:06:35 2004 : Auth: Login OK: [00:c0:a8:86:93:3f/00:c0:a8:86:93:3f] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 17:06:35 2004 : Auth: Login OK: [00:c0:a8:86:93:3f/00:c0:a8:86:93:3f] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 17:06:37 2004 : Auth: Login OK: [00:c0:a8:86:93:3f/00:c0:a8:86:93:3f] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 17:06:37 2004 : Auth: Login OK: [00:c0:a8:86:93:3f/00:c0:a8:86:93:3f] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 17:06:38 2004 : Auth: Login OK: [00:c0:a8:86:93:3f/00:c0:a8:86:93:3f] (from client castle port 0 cli 0.0.0.0)
Fri Feb 13 17:06:38 2004 : Auth: Login OK: [00:c0:a8:86:93:3f/00:c0:a8:86:93:3f] (from client castle port 0 cli 0.0.0.0)

[cdavis]

Forgot to mention, this is a strictly ethernet network. No wireless being used.

[tony]

We appreciate the report. This has been one issue we've been unable to reproduce. Your logs will help, thanks.

What authentication methods are your users using?

[cdavis]

auth mac before login
check framed ip (but only turned that on trying to fix it)
had ez client setup, then turned off and still didnt fix it

you guys want my full logs say so, if you want access to box say so (as I pray a little prayer I don't forget and post access info to the forum :). it is at 100% again. It seems to reproduce itself after every reboot.

[tony]

Please email us with the details. Radius logs will be very helpful as well.

Thanks!

[georgew]

This same problem exist in the most basic set-up. Nat/no nat, no hotspot, simple routing, with wep.

On one machine I see I had remote syslogging on, but it was pointed at my desktop, which did not have kiwi running... but oher machines that have done this were not logging, or were logging to a host that was up... so it's not syslog related. The machines that do this are all under really light load, idle 99% of the time.

[tony]

For all having the 100% symptom, can you let us know if the latest v1.13.4b2 helps.

Thanks!

[cdavis]

Any status on this? Do you guys know of any way to make things work until they are fixed? If there might be a wait please let me know as I will switch back to straight DHCP and write several firewall rules.

Thanks

[tony]

We are still investigating the problem, however we are getting close.

cdavis. The problem you are seeing is indeed related to hotspot. Can you try something for us.

Can you uncheck the radius option from the dhcp auto-auth, and only use the hotspot webpage-based login method and report if you still get a 100% reading sustained for over 10 seconds. After you report back to us, you can re-enable the radius under auto-auth.

You can get rid of the 100% by disabling hotspot, then re-enabling it. (quicker than rebooting).

Thanks!

[cdavis]

Sure, but I am only using MAC authentication so will my users be unable to get online for an extended period of time? Can I set 0.0.0.0/0 in my whitelist and allow everyone to access everything while we try to figure this out?

I did uncheck radius, save settings, apply settings and left it like that for about 2 mins. Radius logs showed no more requests but it still showed 100% the entire time. I then re-enabled radius because of the above mentioned worry.

Carl

[tony]

You can continue to mac-based authetication via the webpage (you have it enabled), so it should continue to work for your customers given they open their browser to do an automatic login before they gain access to the network. The only downside during this test is that your customers will not gain an IP via the radius server, and instead will be issued a dynamic.

Also, what version are you running?

Thanks!

[cdavis]

v1.13.3 3993

I have made the change, disabled hotspot, enabled hotspot. Showing 30-50% but not 100% so far. I will keep an eye on it and let you know. How long do you want me to run it like this?

[cdavis]

Just noticed something that may be of value. Used memory jumps to 42, Free and total free drop to 17, then back to 41 and 18. Like it does when you stop and start a service. I saw it do it about 5 times and not haven't seen it again.


After typing that:
CPU load went higher and higher slowly until it reached 100% and held for at least 10 seconds, then dropped again for about 5 to ~88, now back up to 100% for 10 seconds then dropped to 77 and climbed back to 100 for 10 seconds over and over again apparently.

[cdavis]

It has now held at 100% for well over 10 minutes. I rechecked use radius and disabled, renabled hotspot. CPU load climbed again and now steady at 100%

[cdavis]

I received a new signup as a result of turning off use radius. This user was already in radius and should have authenticated anyway when opening a web page?

[tony]

Thank you. Let me know if it sticks at 100% for an extended period of time (1 minute or more).

Thanks!

Just noticed something that may be of value. Used memory jumps to 42, Free and total free drop to 17, then back to 41 and 18. Like it does when you stop and start a service. I saw it do it about 5 times and not haven't seen it again.


After typing that:
CPU load went higher and higher slowly until it reached 100% and held for at least 10 seconds, then dropped again for about 5 to ~88, now back up to 100% for 10 seconds then dropped to 77 and climbed back to 100 for 10 seconds over and over again apparently.

[cdavis]

Did you change anything? I logged in just now and saw it was not at 100% and I haven't logged into the box and ever seen it at less than 100 since I enabled hotspot.

Carl

[tony]

We have not done anything except advise you of a few things to try. Do you still have auto-auth radius support disabled?

[cdavis]

Yes, I renenabled it right after the Wed Feb 18, 2004 11:21 am post. I logged in this morning and it still is NOT operating at 100%

Carl


It has now held at 100% for well over 10 minutes. I rechecked use radius and disabled, renabled hotspot. CPU load climbed again and now steady at 100%

[cdavis]

I logged in just now and it was pegged at 100 again. I have disabled hotspot on that interface and re-enabled dhcp as I have had several people calling since I turned on hotspot saying they had lost their connection.

[cdavis]

Could this have anything to do with my radius server being external to hotspot?

[tony]

No, external radius servers should not effect anything.

To help us troubleshoot your specific problem, can you upgrade to the latest version (v1.13.4b3), and use the starutil -support command which will generate a file you can email us.

If possible, please do the starutil -support command when your server is pegged at 100% for a period of time.

Thanks!

[cdavis]

I have upgraded to the latest beta and I have not seen the proc pegged at 100% since, however, I just took two calls from two different users that were not able to get online. They both had 0.0.0.0 as their IP. Tailing the radius log I saw Login OK from both MAC's but constantly looping, like station server was asking the radius server over and over? As soon as I rebooted the server and had them ipconfig /renew they obtained the correct IP and everything worked fine.