I have a 7-disk array with a mixture of drives ranging from 2TB to 10TBs for a total of 26TB.   I have a history of drives becoming disabled/emulated. 

During a recent parity check, drive 2 did this. It shows 2048 errors and the parity check was stopped.   I canceled the parity check.

I ran an extended SMART test on drive 2 and it shows no errors.

In the past, JorgeB has been able to diagnose that at least one drive had a bad sata/power connection.  He was able to ID the exact drive.   I suspect that is the case with this recent problem.   Diagnostis attached.

I plan to do the following to recover:

1.      shutdown the array,

2.      removed the disabled drive,

3.      start up the array to make sure the drive was removed.

4.      shutdown the array again,

5.      add the drive back with a different sata cable

Does that sound correct?

My big issue is the INSTABILITY of my array.  Why does this keep happening?  I am very careful not to disturb the server once it is up and running well but it still breaks with a disabled disk every few weeks.  I know that UNRAID is stable so it is frustrating to keep seeing this.

I have a relatively cheap 8i HBA card (about $35) and relatively cheap SAS to sata connectors.  I bought new connecters that cost about 3x the old ones.   Do you think it is worth replacing the current cables as the first step in making the build stable?   

I understand that UNRAID uses the serial number of the drive to assign them in an array so placement of the sas cables should not be critical.   Did I get that right?

If replacing the SAS to sata cables with more expensive cables does not work, my next step would be to replace the 8i HBA card.  I am willing to spend a hundred bucks or so if it would mean a stable build.  Any recommendations on such a PCI-e card would be most appreciated.

plexnas-diagnostics-20250104-1111.zip

I have a 7-disk array with a mixture of drives ranging from 2TB to 10TBs for a total of 26TB.   I have a history of drives becoming disabled/emulated. 

During a recent parity check, drive 2 did this. It shows 2048 errors and the parity check was stopped.   I canceled the parity check.

I ran an extended SMART test on drive 2 and it shows no errors.

In the past, JorgeB has been able to diagnose that at least one drive had a bad sata/power connection.  He was able to ID the exact drive.   I suspect that is the case with this recent problem.   Diagnostis attached.

I plan to do the following to recover:

1.      shutdown the array,

2.      removed the disabled drive,

3.      start up the array to make sure the drive was removed.

4.      shutdown the array again,

5.      add the drive back with a different sata cable

Does that sound correct?

My big issue is the INSTABILITY of my array.  Why does this keep happening?  I am very careful not to disturb the server once it is up and running well but it still breaks with a disabled disk every few weeks.  I know that UNRAID is stable so it is frustrating to keep seeing this.

I have a relatively cheap 8i HBA card (about $35) and relatively cheap SAS to sata connectors.  I bought new connecters that cost about 3x the old ones.   Do you think it is worth replacing the current cables as the first step in making the build stable?   

I understand that UNRAID uses the serial number of the drive to assign them in an array so placement of the sas cables should not be critical.   Did I get that right?

If replacing the SAS to sata cables with more expensive cables does not work, my next step would be to replace the 8i HBA card.  I am willing to spend a hundred bucks or so if it would mean a stable build.  Any recommendations on such a PCI-e card would be most appreciated.

plexnas-diagnostics-20250104-1111.zip

You mention changing the SATA cables, but have you done anything about the power side of things.   Power splitters often cause problems, and also are you sure your PSU can cope with the load while all drives are active during a parity chevk.

If your STAT cables are bad, the SMART check may report a UDMA CRC error count. have you got the error count? I have a similar problem that has bothered me for a long time. I had to remove the bad drive and then set up a new config.my 10-disk array now consists of 9 disks.

Disk2 dropped offline, suggesting a power/connection issue, I also see some issues with disk1 at the same time, but that one recovered, do they share something, like a power splitter?

8 hours ago, Unraid_tjy said:

If your STAT cables are bad, the SMART check may report a UDMA CRC error count. have you got the error count? I have a similar problem that has bothered me for a long time. I had to remove the bad drive and then set up a new config.my 10-disk array now consists of 9 disks.

After double-checking the cables, I decided to change the cables on the bad drive. Once I replaced the cables and ran the parity check, the error was gone. Then, I used the suspect cables to connect a new SSD, and UDMA CRC error counts popped up, which had never shown on the 'old bad drive'. Lucky me, I saved my drive; only a single cable needed to be changed.

On 1/5/2025 at 5:27 AM, JorgeB said:

Disk2 dropped offline, suggesting a power/connection issue, I also see some issues with disk1 at the same time, but that one recovered, do they share something, like a power splitter?

They are both on a single power cable serving 5 drives - drives 1 through 5.

On 1/5/2025 at 7:46 AM, Unraid_tjy said:

After double-checking the cables, I decided to change the cables on the bad drive. Once I replaced the cables and ran the parity check, the error was gone. Then, I used the suspect cables to connect a new SSD, and UDMA CRC error counts popped up, which had never shown on the 'old bad drive'. Lucky me, I saved my drive; only a single cable needed to be changed.

Extended SMART on Disk 2 indicate no errors.

On 1/5/2025 at 5:27 AM, JorgeB said:

Disk2 dropped offline, suggesting a power/connection issue, I also see some issues with disk1 at the same time, but that one recovered, do they share something, like a power splitter?

Disk 2 is on a single power cable with 4 other drives:  1 through 5.

I appreciate the responses.  I am going to do the following to get Disk 2 back online:

1.      shutdown the array,
2.      removed the disabled drive,
3.      start up the array to make sure the drive was removed.
4.       shutdown the array again,
5.       add a new drive with a different sata cable
6.       clear the new drive
7.       assign the new drive to Drive 2 slot
8.       start array.

Did I get anything wrong?

 

I am building another unraid server.   Assuming I can get disk 2 back online, I will transfer the important files to the new server.  Next, I will take this original server down, put in new, more expensive, SAS to SATA cables and see what happens.

On 1/4/2025 at 11:19 AM, itimpi said:

You mention changing the SATA cables, but have you done anything about the power side of things.   Power splitters often cause problems, and also are you sure your PSU can cope with the load while all drives are active during a parity chevk.

That is a very interesting point.   The PSU is pretty beefy at 850w (it is a brand name); however, there are 5 disks on one cable from the PSU.  Does that sound like a lot of load?

18 hours ago, demanding-chief3698 said:

That is a very interesting point.   The PSU is pretty beefy at 850w (it is a brand name); however, there are 5 disks on one cable from the PSU.  Does that sound like a lot of load?

Potentially yes. Typically each cable (or connector) on a PSU represents a separate power "rail". While the power supply may total 850w, each rail has a lower limit. Often times the label on the side of the supply will have information of the distribution of power between the rails. I typically spread my drives out between as many drive cables as my psu has available.

Thank you.  Thank you.  I will try this before rebuilding.

On 1/4/2025 at 11:19 AM, itimpi said:

You mention changing the SATA cables, but have you done anything about the power side of things.   Power splitters often cause problems, and also are you sure your PSU can cope with the load while all drives are active during a parity chevk.

 

On 1/7/2025 at 9:30 AM, primeval_god said:

Potentially yes. Typically each cable (or connector) on a PSU represents a separate power "rail". While the power supply may total 850w, each rail has a lower limit. Often times the label on the side of the supply will have information of the distribution of power between the rails. I typically spread my drives out between as many drive cables as my psu has available.

 

On 1/5/2025 at 5:27 AM, JorgeB said:

Disk2 dropped offline, suggesting a power/connection issue, I also see some issues with disk1 at the same time, but that one recovered, do they share something, like a power splitter?

The common theme of you three experts is a suspect sata power line with too many splits.  Well, it turns out my 7-driver array has 5 drives sharing a single PSU cable with 5 connectors (1 x 5).  I took drive 2 out and replaced it with another 2TB RED WD NAS.  I also dug out another modular sata cable for the Corsair RM850x PS, attached it to the PS, and replaced the power connections of Drive2 and 5.   I precleared the drive and built it back into the array with the same sata data cable.  The array came back without errors.   I am guessing that you are all geniuses and this was the reason for my recurring problem; however, I will wait for a few parity checks and a few months before declaring this one SOLVED.   Many thanks for your wonderful help.