Hi folks,

 

i have a red x-ed drive and can not reasign it.

I did some tests with it - no errors found - everything is clear, also no CRC-Error.

Then i cleaned it, did some tests again... but... i can't reassign it? 

Can someone explain why?

Thanks

If you're trying to rebuild to the same disk, you need to unassigned it, start array, stop array, reassign to begin rebuild.

Thanks for the hint - working - but strange, no errors at all... i am confused.

The "failed" disk has only 4000h runtime and as i mentioned, no errors - hmmm 

If you have it post the syslog from when it failed.

Tricky - did some reboots and i think the log is gone, isnt it?

Yes, when there's an issue always grab the diags before rebooting.

Ok thanks - will do this the next time 

So, short feedback: After 10 h rebuild of Disk 10 everything is working - still no errors, neither on the disk nor elsewere.

The Problem is gone like a Ghost...

I must say, this was the first (strange) problem since i use my unraid (April 2010).

My server had various changes - 3x hardwarebase and all disks and i had never problems with it - until yesterday 

Hi folks,

 

it happened again - a disk got a red x. Same behavior as before, reasigned it, reconstruct it, all ok so far.

And i changed the SATA-cable as well.

But in this case i was able to capture a log. Maybe the experts can see whats going on.

Thanks for your help 

unraid-diagnostics-20170217-1056.zip

SMART for disk1 looks fine but there were several ATA errors during the rebuild, replace the power cable just to rule it out, but it issues persist it may be a bad disk, healthy SMART does not always mean healthy disk.

SMART for disk1 looks fine but there were several ATA errors during the rebuild, replace the power cable just to rule it out, but it issues persist it may be a bad disk, healthy SMART does not always mean healthy disk.

 

I have seen the errors in the logfile and i already did that. Last time it was the Disk 10. After the rebuild, the disk 10 is running since January without issues. Can it be that the SATA-Connectors on the Mainboard have some issues because both faults happen on MB-SATA Connectors?

I also did some test with this disk 1 (and also before with disk 10)  on my Windows PC with HD Tune PRO and found no issues. All Tests were fine without issues so its really strange.

Possible but probably not the most likely.

Ok, then i will keep an eye on it and see what happens next - Thanks for your help 

Normally these types of issues are traced back to an issue in the connection chain. Locking sata cables are highly reconnected. If a non locking cable has little or no friction, you should replace it. If there is tension on a cable, relieve it. Honestly, the cabling infrastructure of your server is the most important thing to a stable array. The gentlest nudge can knock a connector askew by a few degrees, and several hundred billion I/Os later, you have a signal loss and boom, a disk is kicked. Fixing one creates another. Locking cables and drive cages are your greatest protection - locking cables for secure connections, drive cages to keep fat fingers out of the case when doing the most normal server astounds - adding or swapping disks. I've gone years without opening my server once it is stable, yet switched disks in and out at least a size times with zero kicks - except one - and it really was a failing disk.

Normally these types of issues are traced back to an issue in the connection chain. Locking sata cables are highly reconnected. If a non locking cable has little or no friction, you should replace it. If there is tension on a cable, relieve it. Honestly, the cabling infrastructure of your server is the most important thing to a stable array. The gentlest nudge can knock a connector askew by a few degrees, and several hundred billion I/Os later, you have a signal loss and boom, a disk is kicked. Fixing one creates another. Locking cables and drive cages are your greatest protection - locking cables for secure connections, drive cages to keep fat fingers out of the case when doing the most normal server astounds - adding or swapping disks. I've gone years without opening my server once it is stable, yet switched disks in and out at least a size times with zero kicks - except one - and it really was a failing disk.

 

Be careful with the use of locking SATA cables.  WD changed their connector design a couple of years ago and eliminated the part of the connector on the drive that the locking portion of those cables used to lock against.  This also eliminated the force which pushed the two mating connecting surfaces together.  The non-locking connector has a plastic nub on it which provides this force which the locking cable does not have.  If you are going to use locking SATA cables, be sure to test to see that the cable has actually locked onto the drive.  If it slides off easily, use a non-locking cable!

 

There is a thread somewhere on this site about this situation.  And I just found this:

 

  http://support.wdc.com/knowledgebase/answer.aspx?ID=10477

Normally these types of issues are traced back to an issue in the connection chain. Locking sata cables are highly reconnected. If a non locking cable has little or no friction, you should replace it. If there is tension on a cable, relieve it. Honestly, the cabling infrastructure of your server is the most important thing to a stable array. The gentlest nudge can knock a connector askew by a few degrees, and several hundred billion I/Os later, you have a signal loss and boom, a disk is kicked. Fixing one creates another. Locking cables and drive cages are your greatest protection - locking cables for secure connections, drive cages to keep fat fingers out of the case when doing the most normal server astounds - adding or swapping disks. I've gone years without opening my server once it is stable, yet switched disks in and out at least a size times with zero kicks - except one - and it really was a failing disk.

 

Be careful with the use of locking SATA cables.  WD changed their connector design a couple of years ago and eliminated the part of the connector on the drive that the locking portion of those cables used to lock against.  This also eliminated the force which pushed the two mating connecting surfaces together.  The non-locking connector has a plastic nub on it which provides this force which the locking cable does not have.  If you are going to use locking SATA cables, be sure to test to see that the cable has actually locked onto the drive.  If it slides off easily, use a non-locking cable!

 

There is a thread somewhere on this site about this situation.  And I just found this:

 

  http://support.wdc.com/knowledgebase/answer.aspx?ID=10477

 

I am using REDs and GREENs from WD and they all have this Latches. All my Cables have metal latches and fit perfect at the HDDs. The only strange thing is,

that not all Cables sit strong at the Disk - some have wiggle room (left/right) and some sit very tight.

The Cable on Disk 1 which fails has this obscure wiggle room (but with metal latch), so i changed it to a tight on - will see, if this helps.

Sounds good to me.

 

I've been here for close to 10 years. This has always been the Achilles heel. People want to build a server fast and cheap and believe that unRaid protects.then.

 

Those that take the server build and cabling seriously, like you, have problems rarely (like you!) Others, like me, allow our shins to get kicked over and over until we finally take the hint and get shin guards that work. 

Sounds good to me.

 

I've been here for close to 10 years. This has always been the Achilles heel. People want to build a server fast and cheap and believe that unRaid protects.then.

 

Those that take the server build and cabling seriously, like you, have problems rarely (like you!) Others, like me, allow our shins to get kicked over and over until we finally take the hint and get shin guards that work. 

 

In my case, the server is 7 years old and had 3 hardware changes, but this problem with the SATAs is new for me - never had issues with the cables...

Normally these types of issues are traced back to an issue in the connection chain. Locking sata cables are highly reconnected. If a non locking cable has little or no friction, you should replace it. If there is tension on a cable, relieve it. Honestly, the cabling infrastructure of your server is the most important thing to a stable array. The gentlest nudge can knock a connector askew by a few degrees, and several hundred billion I/Os later, you have a signal loss and boom, a disk is kicked. Fixing one creates another. Locking cables and drive cages are your greatest protection - locking cables for secure connections, drive cages to keep fat fingers out of the case when doing the most normal server astounds - adding or swapping disks. I've gone years without opening my server once it is stable, yet switched disks in and out at least a size times with zero kicks - except one - and it really was a failing disk.

 

Be careful with the use of locking SATA cables.  WD changed their connector design a couple of years ago and eliminated the part of the connector on the drive that the locking portion of those cables used to lock against.  This also eliminated the force which pushed the two mating connecting surfaces together.  The non-locking connector has a plastic nub on it which provides this force which the locking cable does not have.  If you are going to use locking SATA cables, be sure to test to see that the cable has actually locked onto the drive.  If it slides off easily, use a non-locking cable!

 

There is a thread somewhere on this site about this situation.  And I just found this:

 

  http://support.wdc.com/knowledgebase/answer.aspx?ID=10477

 

I am using REDs and GREENs from WD and they all have this Latches. All my Cables have metal latches and fit perfect at the HDDs. The only strange thing is,

that not all Cables sit strong at the Disk - some have wiggle room (left/right) and some sit very tight.

The Cable on Disk 1 which fails has this obscure wiggle room (but with metal latch), so i changed it to a tight on - will see, if this helps.

 

I have the feeling that some of the metal latching cables have the plastic nub in addition to the metal locking tab but a lot of them don't.  If you pull on the ones without the metal locking device, you can feel some definite (but light--very light in my opinion) resistance as you pull them off. That is the friction fit provided by the nub.  Without that nub and its friction fit, there is nothing to hold the contacting surfaces of the drive and the cable together.  The metal locking tab can provide that force but it has to have that exact piece of plastic to lock onto (and push against) that WD eliminated.

 

BTW, I consider that the entire SATA drive connector system to be a poster child for the way NOT to design a connector system!!!

BTW, I consider that the entire SATA drive connector system to be a poster child for the way NOT to design a connector system!!!

 

I agree. It's truly awful and the generally stiff cables don't help, either.

 

I wish there was an excellent drive cage with SAS inputs. Problem is I'd want 5 drives in my cage, so it'd need one SAS and 1 SATA. You could run SAS connectors from the controller cards and SATA cables from the MB. Would actually work pretty well in practice.

 

Don't know of anything like that on the market. But that would be much better than all those SAS to SATA breakout connectors. Once you connect SAS, which can be a little tricky, but once you've got it locked - it is NOT coming loose and you are NOT loosing signal.