1.3.2 randomly reboots

[BikeMike]

Hi all,
I've noticed my router randomly rebooting, up to several times per day. This has only happened since upgrading to 1.3.2. It's possibly related to web monitoring, as it hasn't happened in the last 2 days since I disabled that. I had web monitoring set to 100 sites and 100 search queries, and several IP addresses excluded from monitoring.

Hardware is Asus WL500GPv2 (broadcom).
If you need any other info let me know.

[florachan]

I have experience random rebooting too ....

I am on 1.3.1

don't have this problem on the stable branch

thanks

[DoesItMatter]

1.3.2 ran stable for me for days on a WRT54G-TM

No reboots, no instability

HOWEVER

1) I don't use website monitoring

2) I don't use QoS

Florachan - are you monitoring websites? are you on a broadcom?

Have to find a common thread to see what's triggering the reboots.

[BikeMike]

I have QOS on, but web monitoring still off. Uptime is now 3 days. If it's still going ok tomorrow i will switch web monitoring back on and see what happens.

Also I initially had configuration loaded from 1.2 but when the rebooting started I reflashed and reconfigured from scratch, but that didn't fix it.

[BikeMike]

After 4 days uptime with no problems I switched web monitoring back on, no other changes. Within a few hours it rebooted again. I've turned web monitoring off again, as the family is getting a little annoyed

[Eric]

Yeah... I've been testing and I can confirm what you say, though it seems to be taking a couple of days before it dies for me.

This is proving to be a very tricky one to fix -- the problem is that there's nothing I can do to immediately replicate the problem. I've already tried debugging a few of the components by themselves but so far, I haven't figured out what's going on. I don't think it's a memory leak, but somewhere I think data is being written to a NULL pointer or outside the bounds of an array... I just can't figure out where despite several hours of testing.

If I put up a couple of versions to test in this thread, can people install them, activate the web monitor (note -- on some of these test versions, no domains/searches will be recorded, because of code that I've taken out to narrow down the problem), and then report which version they tried and if that version eventually fails? This could speed up figuring out exactly what portion of the system is causing the problem, because right now I'm scratching my head.

I did something like this last year with a bug in the madwifi driver, and it seemed to work pretty well.

[Eric]

Ah... turns out there shouldn't be a need for test builds. I think I've identified and fixed the problem -- it will be included in the next release.