Merlin 386.4 on RT-AC68U no space left with kernel errors every week

[automaton]

As the title, I have a RT-AC68U running the latest Merlin 386.4. It's a fairly basic setup with most things disabled including uPNP (HW acceleration is enabled).

About once a week I begin to have issues with the router connecting and have to reset it, sometimes physically (the 'Reboot' icon in the UI sometimes doesn't work). Here's part the log when this happens:

Jan 30 12:02:04 rc_service: check_watchdog 834:notify_rc restart_watchdog
Jan 30 12:02:08 dnsmasq-dhcp[820]: failed to write /var/lib/misc/dnsmasq.leases: No space left on device (retry in 60 s)
Jan 30 12:02:13 kernel: hotplug2/32594: potentially unexpected fatal signal 11.
Jan 30 12:02:13 kernel: Pid: 32594, comm:             hotplug2
Jan 30 12:02:13 kernel: CPU: 1    Tainted: P             (2.6.36.4brcmarm #1)
Jan 30 12:02:13 kernel: PC is at 0x40206f30
Jan 30 12:02:13 kernel: LR is at 0x40206f08
Jan 30 12:02:13 kernel: pc : [<40206f30>]    lr : [<40206f08>]    psr: 20000010
Jan 30 12:02:13 kernel: sp : bee5aaa0  ip : 00000000  fp : bee5acf8
Jan 30 12:02:13 kernel: r10: 00000006  r9 : 00000000  r8 : 00000000
Jan 30 12:02:13 kernel: r7 : 00000000  r6 : 00000006  r5 : 00000000  r4 : 00000000
Jan 30 12:02:13 kernel: r3 : 000000a0  r2 : 00000000  r1 : 00000000  r0 : bee5ab34
Jan 30 12:02:13 kernel: Flags: nzCv  IRQs on  FIQs on  Mode USER_32  ISA ARM  Segment user
Jan 30 12:02:13 kernel: Control: 10c53c7d  Table: 9a40804a  DAC: 00000015

I don't use the JFFS at all (that I know of), nothing custom. I don't really have that many devices on WiFi either, usually around 30.

Any idea what is going on here? Thanks.

 

[ColinTaylor]

This is the same problem you reported before here. Your router has run out of memory. In your other post you suggested it coincided with you running an rsync backup. The kernel crash is in hotplug2 which is associated with the USB drive.

 

[automaton]

Hi, yes that's true it's similar although the error is different so I thought maybe it was a different error.

As a note though I stopped running the backup since then, and haven't run it for a long time now so this can't be from that anymore. I was hoping no longer running the backup would solve this.

 

[ColinTaylor]

There was another out-of-memory problem reported here. In that case it appeared to be caused by a memory leak in httpd. The OP never posted a follow up so I don't know whether he found a permanent solution.

 

[automaton]

Doesn't seem like they solved it. Also seems like there's a bug somewhere.