SM X11SSL-CF + Crucial 4x16GB CT16G4WFD8213: Crashing

[brumnas]

For anybody considering SM-X11 + Crucial-ECC:

I'm running SM+Crucial since summer now and it keeps sometimes crashing - not often, but ugly enough. It just reboots with no log entries. After the automatically following reboot is triggered another bug, this time SM BIOS, telling you "No RAM modules installed". I thought first that the "No RAM installed" is the cause of the FreeNAS crash somehow - but it's not. All the full MemTests were ok.

After a emailing with the SM support, it seems that the current SM BIOS (2.17.1254, 2016) has a bug in RAM detection after a warm reboot. And this warm-reboot is called after FreeNAS 9.10 crashes. So it seems to be quite unrelated: the FreeNAS crash bug only triggers the SM BIOS RAM detection bug - although it seems to be vice versa as some "RAM-MB combination problem => FreeNAS crash => you see, even POST can't see your bloody RAM". But it's not like that :o).

The support guy sent me a RC BIOS, but I somehow hesitate to burn it onto the board, until there is a final release and may be some patches/updates to the drivers. For now, I conclude it like this:
1. X11 is ok with Crucial ECC, although the MB after a warm-reset plays crazy
2. FreeNAS crashes sometimes, I can imagine it could be as of my second pool, where I mix 2x WD REDs with 1x ST (it's just "another free space" for me to use the unused disks I had around - if it goes, it's ok to me, but it shouldn't bring down the server!)
3. The soon coming SM BIOS2 will fix the dreadful "No RAM installed" message/beeping, which is very unpleasant after a FreeNAS crash, but should be harmless in reality

Anyway, FreeNAS should not crash - but as I'm not sure what is causing this, I can't fill in a bug.

Hope this helps somebody,
Andrej

 

[Ericloewe]

if it goes, it's ok to me, but it shouldn't bring down the server!

It's a minor design oversight in FreeNAS 8-9. All disks contain swap partitions - if they're gone and there's something there, the kernel's pager panics because memory contents were lost. FreeNAS 10 will use mirrored swap to mitigate this.

3. The soon coming SM BIOS2 will fix the dreadful "No RAM installed" message/beeping, which is very unpleasant after a FreeNAS crash, but should be harmless in reality

I've had my X11SSM-F for close to a year now. The BIOS is dreadful - it's probably AMI's fault, since ASRock and Asus have similar issues - and it hasn't seen a single update since release! The list of bugs I've run into is longer than the changelog on many BIOS releases - think of the bugs I didn't spot.

 

[brumnas]

Hm, didn't know that; for me, "Supermicro" was something "super" ;-D. Now I realize it has super bugs; the guy was really friendly and fast responding, but it made me worried as he replied that may be the update he sent me has some more bugs, which is nothing to be concerned about :o).. As you can imagine, I _am_ a bit concerned about burning "a few fixes and a few more new bugs" onto a productive system.. Hm..

Ok, so you are telling me that an rather old ST (2012? FreeNAS originally told me I have to upgrade FW, which was a PITA as the new SM X11 board can't boot into that obsolete IDE compatibility mode anymore) could "sometimes" fail writing to the swap partition? Now of course that could be a problem. This old ST drive is a part of zpool (3 disks together) - I thought somehow the swap is over that zpool but didn't check this; yeah, when the swap is a separate partition and the drive/r fails when writing onto it, then good night.

The dangerous thing is, that it completely looks like a RAM/MB failure, as after the panic reboot the server beeps ugly telling you it could not find RAM, so you get then some real "panic" and memtest and google and find nothing. Annoying.

 

[Ericloewe]

could "sometimes" fail writing to the swap partition

Only if something is wrong, such as a failing disk.

The dangerous thing is, that it completely looks like a RAM/MB failure, as after the panic reboot the server beeps ugly telling you it could not find RAM, so you get then some real "panic" and memtest and google and find nothing. Annoying.

Meh, it happened to me once. It was scary at first, but didn't last long.

The real issue is "why is FreeNAS panicking so much". I'd start by checking the logs for clues.