CRASH when backing up router

[wesleyh]

I have 3 routers with Version 1.4.2 on it. they are configured for AP mode with a static IP address. when i click the backup config button the device crashes and will keep rebooting and the run through a continuios reboot cycle until the router is reset to defaults. I have tried this on 3 seperate units and they are a WNDR3700V2

can anyone tell me what is going on?

[z21]

I just had almost exactly the same problem on a TP-Link WR1043ND with 1.4.2, it would crash when I changed the firewall restrictions config or tried to do a backup, then it would be stuck in a boot/reboot loop. It stayed up long enough to let me get an SSH login and access one or two config web pages, then it would reboot.

Unplugging the WAN interface stopped it crashing spontaneously. If I left the WAN unplugged and didn't use the web interface it stayed up, for 15 minutes at least. If I ran the backup command from the command line (create_backup.sh) it would crash. I'm thinking it is some sort of filesystem corruption or resource exhaustion problem (the WAN interface would cause it to try to write to the bandwidth logs I guess).

I tried logread -f to see if I could see any message about the crash. The last lines were always:

Code: Select all

Oct 22 20:21:01 Gargoyle cron.err crond[2646]: USER root pid 2693 cmd /usr/bin/set_kernel_timezone >
/dev/null 2>&1
Oct 22 20:21:08 Gargoyle user.warn kernel: ipt_bandwidth: timezone shift of 780 minutes detected, ad
justing
Oct 22 20:21:08 Gargoyle user.warn kernel:                old minutes west=0, new minutes west=-780

I reflashed it to the 1.4.2 firmware (perhaps not the brightest idea given that it was crashing!) and it came back up OK, restored the config and have not had any more troubles so far.

[Eric]

First thanks for reporting this -- especially the log output by z21, which suggests that some rebooting problems may be caused by the bandwidth monitor, and not issues with wireless drivers as I had originally suspected.

However, I am having some trouble reproducing this issue,and any further information you can provide about what configurations can cause this problem would be helpful. In addition to what configuration you have set for restrictions and quotas I would be interested to know what time zone you are located in (or at least, what time zone you have set the router to monitor).

The interesting part of the log below is that the router thinks there is a 13 hour time-zone shift going on, which seems odd, and is probably incorrect. So, I'm particularly interested in where you are and what time-zone you have the router set to use.

[hnl_dk]

This part happens here too (every ten minutes) on both my gargoyle routers/AP.

Code: Select all

10-22-2011	18:51:06	Cron.Error	192.168.254.5	Oct 22 18:51:01 crond[1732]: USER root pid 27995 cmd /usr/bin/set_kernel_timezone >/dev/null 2>&1

both are set to "UTC+01:00 France, Germany, Italy".

[Eric]

hnl_dk: That's normal, not an error. Cron is set to run that command every ten minutes, and I believe cron is configured to dump a report of all commands to "error" for some reason, even though it's just logging that the command is run.

So.. no, the set_kernel_timezone is normal. It's the timezone shift, specifically a timezone shift of 13 hours, that concerns me.

[hnl_dk]

Thought so, just wanted to make a note, just in case

[Eric]

One more question for anyone experiencing this problem: Is the web usage monitor active?

I just found a bug that can (sometimes) cause the router to crash when restarting the web monitor, which is also a potential source of this error. I'm trying to track down the source of this problem now.

[z21]

Thanks for looking into this. Yes, the web usage monitor was active. My timezone is NZDT (edit: UTC+13) so it is the right amount of shift, not sure why it would need to do it again after startup though.

The firewall config was 9 access restrictions and two exceptions (the exceptions weren't enabled). I did try editing /etc/config/firewall with vi to remove all the rules but it didn't stop the crashing. After reflashing I restored the firewall config and added the rules I was going to add when it crashed, and it seems to be stable with them in place.

Edit: also QoS for download was on when it crashed initially. In the config I restored when I was trying to get it to work QoS was turned off, but it still crashed.

[wesleyh]

I do not have the web usage monitor enabled unless it is default on the WNDR3700V2 it is using the LAN port to connect to my network and is on just a wireless AP mode only my will not stay up long enough to get into the logs, i have to reset it to defaults to get it back and then reconfigure it.