View Full Version : Hotspot authentication
keith.yoder
12-13-2003, 04:53 AM
We are using the hotspot feature of Star-OS 1.13.2 server edition to authenticate users on our lan. We don't actually have any wireless cards in the Star-OS box. From time to time (usually at least once a day) authentication stops working. We reset the star box and it works again. Even when authentication isn't working, users who are already logged in continue as normal. We use an external radius server which also authenticates our dial-up customers and it also continues fine. It simply receives no packets from the star-os box. I'm trying to figure out if some specific event causes this but so far I haven't had any luck. Any ideas?
Keith Yoder
Oxente.net
Pesqueira - Pernambuco - Brasil
lonnie
12-13-2003, 08:57 AM
What system are you using - CPU, ram, Ethernet, etc?
Are you bridging? Does the Ethernet quit or just the authentication, meaning was traffic still moving throuhg the system?
How many users are you using? Do you see anything that is happenning at a certain number of users?
keith.yoder
12-13-2003, 09:47 AM
CPU is a K6 - 450MHz, 128mb of ram. Right now we have about 30 clients max. Ethernet never quits, just authentication. It seems like after some failed logins it happens more frequently. I'm still trying to isolate it.
I would highly recommend one of our more recent v1.13.3 beta releases which solves many problems since the v1.13.2 release. The final v1.13.3 version is due out in the near future as we are nearing the end of the BETA cycle.
Thanks!
keith.yoder
12-13-2003, 02:19 PM
Will do. I'll get back to you probably on Monday. Thanks.
keith.yoder
12-15-2003, 10:41 AM
Yesterday I upgraded to Star-OS 1.13.3b10. We've been up for more than 24 hours and so far, so good. It seems the problem of losing authentication has been fixed. I'll keep an eye on things here. Thanks for a great product and great support!
Thanks, please keep us informed!
keith.yoder
12-30-2003, 02:43 PM
Our StarOS (1.13.3b10) has started acting up again. This time I was able to log in via SSH during the authentication lock up and verify a few things. The CPU load number said 100% but I was able to get around the interface ok. I suspect the CPU usage at 100% is causing our problem.
Any ideas? Our setup is listed in an earlier post and I think we should have adequate hardware for our network. Any kind of denial of service that could make this happen? Everything will work wonderfully for days and then, suddenly, no one authenticates.
lonnie
12-30-2003, 05:46 PM
This is proving to be a tough one. We have seen the symptoms of 100% usage but we cannot duplicate what triggers it, so we must wait and watch.
keith.yoder
12-30-2003, 06:02 PM
Anything I can do to help? Anything to test?
lonnie
12-30-2003, 06:09 PM
If you find a situation that can be duplicated and that will trigger the high CPU usa we would be quite happy. If we can find what causes it then we can find out what it is doing and then, finally we can fix it.
georgew
12-31-2003, 01:16 PM
When you are at 100% cpu, do a tcpdump on each interface.... see if it is a DOS attack or something...
keith.yoder
01-04-2004, 03:27 AM
We've been running at 100% (or so it says) for about 4 days now without problems. I did a tcpdump and found that we've got a lot of people using our the dns server that are not our customers. Is there a way I can make this available for just our customers?
CompuTron
01-04-2004, 09:22 AM
We've been running at 100% (or so it says) for about 4 days now without problems. I did a tcpdump and found that we've got a lot of people using our the dns server that are not our customers. Is there a way I can make this available for just our customers?
Block DNS requests on your border router that are inbound from the internet.
olebno
01-24-2004, 04:43 AM
Hi!
I have a couple of routers going to 100% CPU also. When they have been running for some hours they go to 100% CPU and more or less stays there. The load fluctuates from 50 - 100% with most of its time it is at 100%. Thos are Pentium 200MHz based IBM computers with 64MB RAM and running as firewall at clients with 1Mbit radio based internet feed.
They work as normal but are really sluggish to manage due to the 100% CPU situation. When running TCP dump, I dont see much traffic on any of the interfaces, but I do see a lot of ARP'ing on the external interface. Someone in the same segment probably has a worm of some sort?
When i need to do management on those boxes, i log in and do a Activate Changes form the menu. Then after very short time, the unit gets back to 1 % CPU and stays down in the normal CPU load for at least a couple of hours.
Best Regards
Oleb
keith.yoder
01-31-2004, 01:41 PM
We are running the last beta of 1.13.3 (11, I think) and still getting 100% cpu usage and hotspot lockups. The only other thing I've noticed is that I can't access the radius configuration webpage or other webpages when this happens. Again, users who are already authenticated continue to operate as normal.
lonnie
01-31-2004, 07:56 PM
How many users are on the hotspot system? We have been fighting with this and cannot really duplicate it here.
keith.yoder
02-03-2004, 02:25 PM
We have about 30 users on the hotspot system.