N100 corruption on updates (ZFS)

[sarkyscouser]

I have an issue which may be related to: https://forum.opnsense.org/index.php?topic=48343.15 but I'm using the ZFS file system and I've implemented the tunables recommended in that thread and rebooted before I upgraded (as I suffered this issue last upgrade last week). 

When I update (as I just have from 25.7.4 to 25.7.5) I'm experiencing an issue where the update doesn't work first time and my Adguard Home and Unbound config reverts to a previous version and I have to update again (which then works) and restore my Adguard Home and Unbound config. This issue also happened when I upgraded from 25.7.3 to 25.7.4.

I'm using ZFS on a sata ssd and there are no SMART errors and I can't figure out what's going wrong.

I've attached a system.log file from today, but not sure what to look for?

Edit: I'm also now noticing that on just a plain reboot (no update) that my unbound config is reverting to a previous version as well, so something funky is going on

[franco]

I don't see the update data but what you describe sounds more like an issue with third party repos being out of sync and the package manager trying to cope with that fact.

Ideally, all third party repos should be rebuilt after each OPNsense stable release due to the fast moving nature of third party dependencies.


Cheers,
Franco

[sarkyscouser]

So run a pkg update after upgrading each time?

[franco]

It's not your fault this happens. You can check for updates and install again, but the third party repos need to be updated so this back and forth doesn't happen anymore. Otherwise it keeps happening.

I wouldn't worry too much about it as long as things keep working either way.


Cheers,
Franco

[Patrick M. Hausen]

If it's only AdGuard Home you are running from @mimugmail's repo - he published a second repo containing only that package in December last year to minimise potential conflicts:

https://www.routerperformance.net/opnsense-repo/

[sarkyscouser]

Yes that's what I'm running the adguard home only repo.

I'm not convinced this is the issue though.  When I upgraded from 25.7.3 to 25.7.4 I got a "/" corruption error but at the time didn't report it as I was short of time.  I suspect the same has happened again but it's not shown in the logs and I didn't see it on the screen during update.   The SSD seems fine via SMART so like the other thread I linked above there seems to be a possible N series Intel corruption issue lurking somewhere?

[meyergru]

You should be generally safe from those problems with ZFS alone and even more so when you apply the settings from this post plus the Intel microcode updates.

That being said, there were reports of defective BIOSes and RAM. Did you run a stability test?

[sarkyscouser]

Hi, I already have the microcode and those tunables applied after I spotted them from last time I had an issue.

I guess I'll see how it goes next set of updates and report back but would be interested to see if anyone else with an N series chip experiences continuing problems.

[Patrick M. Hausen]

The Intel N100 bugs cause random file system corruption, not arbitrary yet valid and consistent changes to an installed system. That's simply not possible.

Your root cause must be something different.

[sarkyscouser]

I've ordered a new ssd and will reinstall next week and then see what happens during the next scheduled update in 1-2 weeks time, not sure what else I can do for now?