Controller issues / drive disappeared

[Kevek79]

20 minutes ago, steve1977 said:

Got it. So, you applied it directly to the drives itself (where the sata cable connects to). First "red", then "gold" right thereafter. No drying or anything, just apply one after the after? It is a new thing that I haven't tried out yet. While I still suspect the over-heating of the controller card to be the primary issue, it's worth to give it a shot to rule out all options. Also, the over-heating may be even more severe since I added a meaningful number of 10Tb drives. I remember that a few years ago, I could solve the issue by only connecting 7 instead of 8 disks (or running two controller cards each with 4 disks). I believe this may have helped to reduce heat?

You can try to mount a fan directly to the heat sink of your controller card to rule out any overheating issues as well.

If the heat sink is the same size than on the LSI cards, this 40mm fan might be of interest to you.

Noctua NF-A4x10

 

All those SAS cards are made for servercases with high airflow, and they dont play well in a standard tower without propper airflow over and around the card. (Especially during summer months )

Edited August 20, 2019 by Kevek79
additional info

[steve1977]

In anticipation of the heater issue being present after controller card change, I have bought the following fan:

 

[steve1977]

In anticipation of the heater issue being present after controller card change, I have bought the following fan. I haven't installed it as it seems more difficult than I anticipated to set it up correctly. A bit more than plug & play

 

https://www.amazon.com/Titan-Adjustable-Dual-Cooler-TTC-SC07TZ/dp/B008A2TDC6/ref=sr_1_1?keywords=Titan+TTC-SC07TZ(RB)+VGA&qid=1566294846&s=gateway&sr=8-1

[itimpi]

10 minutes ago, steve1977 said:

Thanks, very helpful. Have you ever experienced any heat issues with your controller card? Even if just plugged in for a few minutes, it gets incredibly got (if touching the heat sink).

Mine never seems to get to hot to touch so heat does not seem to be a problem for me.   I am using a LSI 9201-16i card.

[steve1977]

I am not sure I'd call it "too hot to touch", but clearly very hot. Maybe I give it another try to use 2 controller cards (each with 4 disks). Not sure though whether Unraid will allow me to rebuild the array after changing controller connections.

[itimpi]

1 hour ago, steve1977 said:

Not sure though whether Unraid will allow me to rebuild the array after changing controller connections.

Unraid does not normally care how the disks are connected - it recognizes the disks by their serial numbers.   The only time you might have any issues is when the controllers involved report the serial numbers in different formats.

[steve1977]

Just added a second controller. Issue remains that disk 12 facing issues. May I ask your help to look inside the log whether any new info?

 

I am getting new cable tomorrow. Maybe this helps.

tower-diagnostics-20190823-1001.zip

[JorgeB]

Not seeing any issues on disk12, I see the same old issues on disk13, and it's still on the same SATA port.

[steve1977]

This is really odd. Disk 12 is clearly the problem child. It either doesn't show up at all. Or it show up as "unassigned". Error messages around drive not being ready. I am starting to think whether I should just replace disk 12?

[JorgeB]

If it doesn't even show up try swapping it around with another disk on a different controller, also swap power cables, if it still doesn't show up

it's likely the disk.

 

Also don't forget disk13, still original problem.

 

 

[steve1977]

Thanks, appreciate your help!

 

Ok. Somehow disk 13 does not make trouble directly, but it could well be the root cause of something else.

 

I have changed the cable of disk 13. Any improvement in the log?

tower-diagnostics-20190823-1054.zip

[steve1977]

Even weirder... Now after the cable swap, it is the first time that disk 13 shows an issue. It shows that disk 13 has no file system. Shall I start in diagnostic mode and try to get it back?

 

And funny enough, it started to rebuild the disk 12. Not very fast, but faster than before... So, it seems the cable change may have helped with something?

[JorgeB]

3 minutes ago, steve1977 said:

it is the first time that disk 13 shows an issue.

No, it was having read errors before, probably why it's unmountable, but the diags are pre-array start.

[steve1977]

I know I know... Sorry for unclear communication. It is the first time disk 13 shows visibly issues. Before. it doesn't show read errors in the GUI and shows as "green".

 

Let me go to maintenance mode and make oit mountable again.

 

Does the latest diagnostic still shows errors? I have changed the cable for disk 13 (but not for any other disk). Brand new cable!

[steve1977]

Here we go with updated diagnostic and also result from running repair -n

 

https://pastebin.com/8v1h6pkJ

tower-diagnostics-20190823-1113.zip

[JorgeB]

2 minutes ago, steve1977 said:

Does the latest diagnostic still shows errors?

Not for disk13, though like mentioned they are pre-array start, but errors were still present on previous diags even before starting the array.

[JorgeB]

1 minute ago, steve1977 said:

also result from running repair -n

Run without -n, will also likely need -L, use it if asked.

[steve1977]

Got it, thanks, done it.

 

There seems some progress... Either because of second controller or because I changed the cable of cable 13.

 

It is still very slow though in rebuilding the parity. May I ask you to have another look into the log file? Are the errors from disk 13 gone? What about other disks or anything else suspicious in the log?

tower-diagnostics-20190823-1125.zip

[JorgeB]

Log is clean, no errors so far, as for the speed there appears to be something else reading from disk1, possibly cache dirs, wait a few minutes and make sure nothing else is accessing the array and speed should improve.

[steve1977]

Got it, thanks!

 

So, it appears that a cable issue of disk 13 caused disk 12 to fail and disappear. Disk 13 itself never showed any read or other errors (except in log). Is this somewhat explainable and makes sense?

 

I remember you'd mentioned disk 9 also showing some issues earlier? Assume it may be worth to change the cable as well?

[JorgeB]

4 minutes ago, steve1977 said:

So, it appears that a cable issue of disk 13 caused disk 12 to fail and disappear.

Unlikely to be related, IIRC disk 12 was on a different controller.

 

4 minutes ago, steve1977 said:

I remember you'd mentioned disk 9 also showing some issues earlier? Assume it may be worth to change the cable as well?

Yes, there were only a few errors, much less than disk13, still worth replacing it.

 

[steve1977]

Thanks. I thought all worked out well since changing the cable of disk 13.

 

Unfortunately, just a few minutes after the array rebuilt completed successfully, disk 12 shows error messages / disabled. Do you see anything in the log? What about disks 9, 12 and 13? Or others?

tower-diagnostics-20190824-1645.zip

[John_M]

I had an LSI controller (on a Dell H310 card) that had problems with overheating. So I removed the heatsink and found that the thermal compound was completely dry. In fact it crumbled into pieces and fell off as I removed the heatsink. I cleaned up what remained with isopropyl alcohol and applied a small amount of Arctic MX-2 before replacing the heatsink and that fixed the problem without the need of an extra fan.

[JorgeB]

ATA errors on disk13 are back, and then disk12, make sure power connectors on those disks are OK, or swap with another.

[steve1977]

Thanks. Let me switch the power and SATAA cable of disk 13 with a neighboring disk. Let's see whether the issue "moves on"... Will report back.