What does a "NOP FLUSHQUEUE" message mean?

[StephenFry]

I'm messing around with a test-system that I have upgraded from 9.10.2-U3 to 11-STABLE.
It uses the same h/w as the main SuperMicro system in my .sig.

It's working absolutely fine, but during boot it says about 3 out of 8 drives "nop flushqueue", error 5, retries exhausted.

For the life of me I can't get this system to fail or otherwise behave badly, but seeing this scroll by during boot is a bit odd and of course I want to try and solve it, in case it is in fact an error.

Knowing where to start solving it means knowing what the error _is_ - and it seems to be very rare indeed. Google is not helpful.

Any gurus here know what it means?

 

[joeschmuck]

11-STABLE

There is no Stable yet, that is the train you are on. Is it 11-RC1 or RC2?

Have you checked the SMART data for the drives in question? Also, are the drives connected via the M1015? Can you take those suspect drives and hook them up to the motherboard directly and retest?

Here is another thread discussing this problem, but no solution so maybe you could ask how it was fixed, assuming it was fixed.
https://forums.freenas.org/index.ph...onnect-to-the-volume-detect-by-freenas.54188/

 

[joeschmuck]

Here is the only reference I could find for this error message.
https://github.com/freebsd/freebsd/blob/master/sys/cam/ata/ata_all.c

 

[Stux]

Looking at the source code, NOP FLUSHQUEUE occurs when the ATA command being executed is 0x00, with 0x00 as the parameter.

The cmd is the cmd parameter for the ATA I/O data structure, and I guess its possible there's a bug causing an empty data structure to be 'executed', thus generating the NOP/FLUSHQUEUE.

Or there is actually a demand for NOP/FlushQueue from somewhere... Would need more debugging to determine what is happening.

 

[StephenFry]

There is no Stable yet, that is the train you are on. Is it 11-RC1 or RC2?

Have you checked the SMART data for the drives in question? Also, are the drives connected via the M1015? Can you take those suspect drives and hook them up to the motherboard directly and retest?

It is the latest, RC2.

Yeah, the drives are perfectly fine. They also do not give any errors when connected to another test system I keep around, but that is on 9.3.1 currently.

After some more testing, and connecting the drives in different configs (M1015, motherboard, etc) I cannot make heads or tails of this. Just now 5 drives gave the NOP FLUSHQUEUE notification, then I mix all the connectors around, and it's all 8. Rebooting -with nothing changed- and BAM!: it's 4 drives. It just seems to be completely random.

And to make things interesting, going back to 9.10.2-U3, which didn't complain before, it too is suddenly NOPping a-go-go. Would I be a normal user and not look at my screen during boot time, I would not know something might be amiss. I've done a (shortish) burn-in test, even though this test system has proven itself, and it all works f.i.n.e.

Flabbergasting, but for now I'm leaving it as it is. Life is too short for troubleshooting trouble that may not even be there. Thanks all for the replies and maybe future me will revisit this thread and think, "Doh! Why didn't I pursue this!". Sorry, future me.

 

[joeschmuck]

Well this is a last ditch effort, how about a clean install of FreeNAS on a new boot device? Also, I'd run Memetest86 and some CPU stress test to see if the hardware is still stable. You could have a power supply going bad.