Jump to content

Unmountable cache pool after reboot


Recommended Posts

Hi guys. This is my first post to the forums in a long time.  I was in the process of coping data from drive to drive and formatting drives with the btrfs as I have a number of drives that were formatted with riserfs.  I moved some data from disk 1 stopped the array, changed the file system for the disk, swapped in a new precleared disk, and started the server. I hadn't touched anything on the cache pool. I think.  When I started the server unRAID was telling me the pool was unmountable because there was no file system on the pool. I pulled the drives that make up the cache pool and plugged them into my desk top to test.  And now after reinstalling them into the server unRAID is telling me there is no pool uuid.


The pool was configured to run RAID 10.  I had Dokers and a VM running on the pool. I have searched a bit but I am in a little over my head at this point. 


I have attached 2 syslogs. One from when the pool was unmountable and another from when the pool is missing the uuid. 


Thank you in advance 

ah poo.PNG



Ah poo 2.PNG

Link to comment

Your first syslog contains no reference to any of your Samsung SSDs. As far as unRAID is concerned they don't exist. I assume they have dropped off line due to a communication failure between the controller and the SSDs - so typically a cable/expander/backplane issue. That syslog does not match the first screenshot though, which shows the SSDs all present. I feel that the screenshot is later along the timeline of events than the syslog.


The second syslog shows the SSDs all present again but unmountable because of the UUID problem. That does match the second screenshot.


Other than that I have no information. You ought to have grabbed diagnostics (Tools -> Diagnostics) when the problem occurred, not after rebooting. Always post the complete zip file. You have a lot of disks and several different file system formats and it's difficult to keep track of the different disks from the syslog alone.


My guess is that at the time of the first syslog you could have recovered by shutting down, checking connections, then restarting in maintenance mode and running a file system check on the cache pool. You said that


2 hours ago, elco1965 said:

I pulled the drives that make up the cache pool and plugged them into my desk top to test.


and I suspect that's what broke the UUID.


Link to comment

I did do a few other things in between.  I did stop the array and check connections. Had the same unmountable  no btrfs problem.  Stopped again unplugged from the card they are plugged into and plugged directly into the MOBO. Had the same unmountable no btrfs problem. 


You are correct, the syslog was saved after i tried to reconnect/different connections with the pool hoping the issue lied within a cable/card issue.


Is it safe to assume that even though the data on the drives should be good, maybe? That I will have to rebuild the cache pool and start from scratch? 

Link to comment

I don't know enough about the subtleties of btrfs to be able to advise you there. It might be possible to create a new UUID without reformatting, for all I know. Since you have invested in btrfs on some of your data disks you might find it useful to read a bit more about it. At best, though, that would get you back to your original position of an unmountable file system, which you might be able to repair or you might not. Whenever you've got an unmountable file system the majority of the data is generally "still good" - it's just that you don't know quite how to access it. If the data is backed up, as it ought to be, then I wouldn't hesitate to start from scratch.

Link to comment

Thank you John. I will educate myself further. 


I just wish I knew what caused this so I can try to prevent it from happening in the future. 


If no one else chimes in I may just rebuild the pool and start from scratch. At some point, I guess the juice just isn't worth the squeezin.

Link to comment


This topic is now archived and is closed to further replies.

  • Create New...