Reboot and lockup issue

[pti-andy]

I am having an issue with V1.1.0 on WAR-2's rebooting several times a day and sometimes locking up. This is a simple point-to-point bridge with Cisco routers on each end. The WAR’s have single CM-9’s with clean 24V-1.5A power. The settings are 48Mbps, 2X cloaking, Super A/G, and AP power save off. I tested the link for several weeks running various file transfers up to 12Mbps with no issues.

Now that it is in production we are averaging about 4.5Mbps and either end can reboot or lockup. I have replaced the WAR’s and still have the problem. The CPU utilization averages from 5-25% but sometimes reaches 100%. The only difference I can see is that now the traffic is made up of our data center hosting which has thousands of destination IP’s where before it was just local traffic. Any ideas?

-Andy
PTI Wireless

[lonnie]

Try and run it without connection tracking and do not do any sort of firewall or bandwidth rules. Turn ON the AP Power saving. If it is talking with our unit it is required and it works properly.

[pti-andy]

I currently do not have any firewall or bandwidth rules. I have noticed that it frequently locks up when trying to login to manage it. The reboots are all on it's own though.

I will turn off the connection tracking and turn ON the AP power saving. Just what does power saving do for a point-to-point link that has bandwidth running 24x7?

Thanks for info.

-Andy
PTI Wireless

[lonnie]

Every Atheros radio must periodically do a recalibration of the RF section and the client must do a background scan to keep track of what is going on around it. It must stop talking with the AP when it does that which would lose data. The idea is to inform the AP before it stops talking and the AP buffers data while the client is doing other things. When the client comes back the AP will then catch up and send the data it has been buffering. You might see a spike in latency but nothing gets lost.

[pti-andy]

Ok, that makes sense. I was thinking it had something to do with power management and didn't want it to go into a low power mode so that's why I had it off.

Thanks for the explination.

-Andy

[pti-andy]

It seems that turning off the connection tracking fixed the self rebooting issue. It has been running clean for three days without a reboot, however; it still does lock up every now and then when logging in to manage it thus requiring a power cycle to bring it back.

I noticed that the quagga-watchguard service is running on some of my V3 units and not on others. The unit that was self rebooting used to have this service active and now doesn't. I made sure that connection tracking was turned off on the units that show this service. What makes this service go away and why does it appear on some but not all of my sites? Is it even needed?

Thanks,

-Andy
PTI Wireless

[pti-andy]

In case my previous post was missed... Can you please explain why the quagga-watchguard service is running on some of my sites but not all when they are all configured with the same settings? What dissabled it?

How can I turn this off if it is not needed for simple bridging?

Could this have been related to the rebooting issues that I discussed in my previous post?

-Andy
PTI Wireless

[lonnie]

Its job is to simply restart Quagga services if they fail. If you do not have RIP or OSPF enabled then there is no reason for the watchdog and it will do nothing. It should be running and we will have to see why it quits.

Check your cabling and power for any reboot issues. Some guys have found squashed or rubbed through cables. When you prepared the cat5 connection did you use the included keystone jack or did you crimpthe RJ45 directly to the solid cable?

[pti-andy]

As stated in a previous post, it no longer reboots since shutting off connection tracking. It still does lock up but only when logging into the unit now. It used to do this when running bandwidth tests from the WAR and once during a flash but I don't do this any more so now it is limited to logging in.

The Quagga watchdog service stopped showing yellow on the main page shortly after shutting off connection tracking. All of my sites that are running heavy bandwidth with a bridge now have this off. Only one site still has it on and it gets very little traffic although it is configured the same as the others. Hope this helps.

-Andy
PTI Wireless