Been having an intermittent issue where Unraid was going unresponsive and I'd have to hard shut-down then power it back up, where it'd run for several days, then go unresponsive again. At the advice of @JorgeB I enabled a remote syslog server and have 'caught the Kernel panic in `kernel.log`. I attached the last few hundred lines that shows the what looks like a General Protection Fault, followed by the Kernel Panic as well as all output.

 

I've already let memtest run for 24 hours on machine, 4 passes and no problems detected. I'm attempting to re-create my flash drive since I saw that mentioned in other posts, but running into the Unraid flash creator can't see the drive bug, but once I get that straightened I'll be running with a new flash to see if that works.

 

I don't know where to go from here in terms of diagnosing and fixing, so would appreciate smarter minds than mine hopefully being able to point me in the right direction.

 

System Specs:

Unraid v6.8.3

Dell R710

x Xeon CPU E5620

32GB Ram

12x 4tB Drives (10x array, 2x parity), 2x 250GB drives in cache pool

kernel.log

Crash appears to be network related, try to simply the network config as much as possible and/or try a different NIC.

thank you for the advice. I will shut off one of the NICs and try to operate with only 1.

I've since realized that the switch I'm using (Cisco SG112-24) is reported by other people as having oddities on other forums, so I'm also sourcing a new switch to use.

@JorgeB Forgot to ask, what specifically did you see in the `kernel.log` that pointed to NEtwork so if it crashes again I can look for it?

The module mentioned in the call trace:

nf_nat_ipv4

 

so, resurrecting this since it seems related. After @JorgeB advice, switched to using single NIC. Had no issues.

In the interm, since it appeared to me that the previous unmanaged switch I was using had issues with lags and bonds, I picked up an Aruba S3500-24P used. Switch appears to be working without issue to the two other servers I have setup with bonded NICs (2x ESXi servers).

I setup a 802.3ad bond between the two NICs on Unraid, and the two ports on the switch. No errors in the log about NICs anymore. Everything has been running fine for ~10 days and change until last night Unraid crashed with another General Protection Fault. Going from the last reply here, the `RIP` says `nf_nat_setup_info+0x365/0x666 [nf_nat]`, so I would guess it's still network related?

 

Anything else I can try to use both NICs on this server? I'd rather not be stuck with a single 1G connection for an Unraid storage server.

unraid-crash-2021-01-03.log

single bump just to see if anyone has any ideas. I have the new NIC installed, just wondering whether I should stop using the onboard NIC altogether, or the crash was related to a software issue. It hasn't crashed since (which is good), but when it does crash I end up having to rebuild my containers because docker crashed and it's a pain.

7 hours ago, Zxurian said:

just wondering whether I should stop using the onboard NIC

I don't think so, I use onboard in active-standby with add-on 10G NIC currently, but some previous build was use onboard only and never have problem.

The different is I never use 802.3ad bond, may be this not robust and cause problem. 802.3ad bond also switch relate but not means switch cause this.