kernel panic during boot

[axelmahone]

My friends freenas does not boot anymore.
We updated his system from 9.x to 11.2 a couple of weeks ago.

A short review over the process:

• we installed a fresh version of Freenas 11.2 on a usb drive
• we restored his ZFS pool (RAIDZ1)
• we devided his data volume into sub datasets and moved the data
• set up users and shares
• he bought a new ssd and we copied the boot device to the new drive (with dd)
• created a jail (on the data-volume)

It was running fine for a couple of weeks, but now he gets an assert (zap_increment_int [0x7a == 0x0]) resulting in a kernel panic while booting the system.



According to this resource it is a checksum error: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=220203
This case was in connection with a SAS controller, which we do not have.

Does anyone have an advice how to fix this issue?
It is for private use and I always have the option to reinstall the complete system within a couple of hours, but out of curiosity I want to try to find and fix the cause.

Hardware:
ASRock 75 Pro+ (latest UEFI)
AMD A4-3400
16 GB RAM
250 GB SSD boot device
4*3 TB WD Red

(no need to point out it's consumer hardware, I am totally aware of that :) )

 

[sretalla]

What brand/model is the boot SSD? if this is one of the drives impacted by the 3D NAND/TRIM issue, you may need to replace it.

Typically this would be WD Green or some Crucial drives, but can be others too.

 

[axelmahone]

It is a SanDisk Plus and it seems to be impacted:
https://redmine.ixsystems.com/issues/35065
Thanks for giving me the hint!

I will try to disable TRIM and see if it will work than. But it looks like the results seems to vary.
I am not sure how to do it yet, but I will figure it out.

 

[sretalla]

Also be aware that disabling TRIM is system-wide, so all SSDs will be impacted by that.

The better way forward is with a $30 Kingston SSD (doesn't need to be very big). Use the SanDisk in a Windows or Linux machine for now until FreeNAS squashes that bug.

 

[Chris Moore]

Intel SSD drives have worked well for me. You can pickup a very good used model for little money:
Something like this would probably outlast two servers:

Intel 320 Series 80GB Internal 2.5" (SSDSA2CW080G3) SSD for sale online | eBay

Find many great new & used options and get the best deals for Intel 320 Series 80GB Internal 2.5" (SSDSA2CW080G3) SSD at the best online prices at eBay! Free shipping for many products!

www.ebay.com

I have some that I have been using for four years with zero trouble.

 

[axelmahone]

Yesterday I did a scrub on the boot pool. Now I can confirm there were checksum errors.
He bought a new USB flash drive today and I will install a fresh copy of FreeNAS.

Is it a bug in FreeNAS or with ZFS in general? I am using some NVMe SSDs with ZFS in my rigs.

 

[anmnz]

Yesterday I did a scrub on the boot pool. Now I can confirm there were checksum errors.
He bought a new USB flash drive today and I will install a fresh copy of FreeNAS.

Is it a bug in FreeNAS or with ZFS in general? I am using some NVMe SSDs with ZFS in my rigs.

The bug appears to manifest with specific SSD controller hardware, which means it is likely a bug in the firmware, or a bad interaction between the firmware and the FreeBSD hardware drivers.