urgent Q ! unable to open enough prompts for badblocks tests

[VladTepes]

OK so I reckon that by the time I get home the badblocks tests will be completed. Hope so at any rate.
How do I then go in and check the results? Is there a command I can use?
I ask because when I look in a tmux session I can see that the info has 'scrolled up' within the area assigned to that disk and so can't all be seen....
(I originally opened a tmux session to start the tests and did crtl B and shift" to split the screen)

 

[danb35]

The last line output by badblocks will say something like "test complete, 0 errors found". If you see that, you can do Ctrl-D to end/close that window and tmux will move to the next one. Repeat as needed.

 

[VladTepes]

So that's per 'window' within tmux, not per session?

 

[danb35]

Correct.

 

[VladTepes]

Badblocks results...

Well I went through the tmux sessions and can't find the badblocks results for drive ada3. I know it was running, but seems to have scrolled off the window.... I'm confused.
(see attached file), I chose the tmux commands for next window etc no luck.

Code:

ada1   
(0/0/0 errors)

ada2
(0/0/0 errors)

ada3
???

ada4
(3/0/0 errors)

ada5
(0/0/0 errors)

ada6
(3/0/0 errors)

In short a couple of the drives show errors 3/0/0, which I assume means 3 read errors. Which I assume is negligible. Correct?

I'll run long smart tests again and see what happens :)

 

[Stux]

Badblocks results...

Well I went through the tmux sessions and can't find the badblocks results for drive ada3. I know it was running, but seems to have scrolled off the window.... I'm confused.
(see attached file), I chose the tmux commands for next window etc no luck.

Code:

ada1 
(0/0/0 errors)

ada2
(0/0/0 errors)

ada3
???

ada4
(3/0/0 errors)

ada5
(0/0/0 errors)

ada6
(3/0/0 errors)

In short a couple of the drives show errors 3/0/0, which I assume means 3 read errors. Which I assume is negligible. Correct?

I'll run long smart tests again and see what happens :)

Are these new drives?

I would've expected zero errors.

Just ran badblocks on 8 new 4TB drives the other day, zero errors. Which as I said is what I expected. Any errors and I would've returned the drives.

Next thing is to verify the drives all have reasonable performance with sooner array tester.

 

[Ericloewe]

SATA can be rather finicky. Just running 0.5m cables with the chassis open in my server is enough to cause interface errors. And no, I'm not sitting next to massive industrial machinery in a dirty EMI environment.

 

[Stux]

Do those errors show up in a badblocks pass?

 

[Ericloewe]

Do those errors show up in a badblocks pass?

Yeah, timeouts mostly. I was rather surprised, but the server passed burn-in with the side panel on, so I just metaphorically signed it off.

 

[VladTepes]

OK so 3 of the drives are new (in warranty) and 3 older (out of warranty). ada4 and ada6 are older drives. Not too many hours on them (around 2,000) but have been lying around a while.

Interestingly at the moment the side panels are off and cable runs are messy. I'm waiting for some new quality custom cables to arrive to tidy it all up.

 

[Stux]

Yeah, timeouts mostly. I was rather surprised, but the server passed burn-in with the side panel on, so I just metaphorically signed it off.

I think that might be why esata cables have extra shielding

 

[Stux]

As long as the failing drives are the ones out of warranty. You should run the smart tests next.

part of running badblocks is to let the drive find bad blocks and stop using them.

Power on hours doesn't effect the rust, but rather the mechanicals. The rust will rot all by itself

If it's just the occasional error, ZFS will take care of it.

 

[Bidule0hm]

SATA can be rather finicky. Just running 0.5m cables with the chassis open in my server is enough to cause interface errors. And no, I'm not sitting next to massive industrial machinery in a dirty EMI environment.

Not normal, I'd say it's crappy cables.

 

[Ericloewe]

Not normal, I'd say it's crappy cables.

More likely overused cables that were still good enough for the casual observer not to notice. Those cables had seen a few setups with... "interesting" bends.

 

[Bidule0hm]

Can be that too :)