I have 2 parity drives and contents are emulated.  Running unraid 6.3.5 with two sgi se3016 (see link https://www.servethehome.com/sgi-rackable-3u-16-bay-se3016-sas-expander-chassis-forum-deal/)

 

In the past i have re precleared the read error drives and most of the time, there is nothing wrong on read errors.  How safe is it to re introduce one or both of these drives to array and how would i go about it.  Also can someone look at smart reports and advise?

Additional info.  When i downloaded diagnostics, it gave me logs that ended on October 20th (today is Nov 7th).  I was unable to get log to show up clicking on the log icon on tower page. 

I could also post the logs that i had that ended October 20th?

tower-smart-20171107-0736.zip

tower-smart-20171107-0735.zip

Disks look fine, two disks getting disable at the same time would point to a controller/power issue or hiccup, but without the diagnostics from when it happen it's just a guess.

 

If both emulated disks show the correct data you can rebuild both to old disks and at the same time:

 

https://wiki.lime-technology.com/Troubleshooting#Re-enable_the_drive

Thanks!!

 

So, if i rebuild 2 drives at the same time, i will not have any drives for parity?  If i lose a drive or have an error during rebuild i have a problem if i understand correctly?  Perhaps rebuild one drive at a time?  My assumption is that both drives with read errors have all the data on them since it was only read errors.  If the system went into emulated drives, is it possible that data was written to the emulated 4 or 5tb?  Seems really unlikely but thought i would ask.  I assume that the emulation would just put the two disabled drives out of the list of drives that can be written to(even virtually). 

 

It appears from the link you posted would allow to re-enable without rebuilding, but is not advised?  I think i am probably going to search for old drives that match the size of read error drives (4 and 5tb) preclear them and then restart array (keeping the 2 read error drives as a backup) 

 

I have 1 hot precleared spare that is 6tb, but i always end up having to get the data off that drive and bring it back as a hot spare in situations like this (once every 8 months or so i get one of these read errors).  Gets pretty time consuming with the drive in error state generally being ok after all is done.  Any advice on how to proceed before i start the journey?

15 minutes ago, FrozenGamer said:

If i lose a drive or have an error during rebuild i have a problem if i understand correctly?

Yes, but the same will happen if you just rebuild one at the time and lose another drive during the first rebuild, if one of the disks being rebuild fails it's not serious as you'll stay as you are now, if a different disk fails it can cause data loss, no matter if you rebuild one or both.

 

16 minutes ago, FrozenGamer said:

If the system went into emulated drives, is it possible that data was written to the emulated 4 or 5tb? 

That only depends you your share allocation and split settings, for writes emulated drives are treated as normal drives by unRAID.

 

18 minutes ago, FrozenGamer said:

It appears from the link you posted would allow to re-enable without rebuilding, but is not advised?

If you sure no data was written to the emulated disks and the disks itself are fine you could do a new config and sync parity, but if one of the disks fails during sync you have no way to rebuild it after.

 

If you sure no data was written to any data disk (including the emulated disks) after they got disable and the disks are fine you could do a new config and trust parity, then you'd need to do a parity sync because at least a few sync errors are expected.

 

 

Thanks Johnnie.  I chose to use a 5tb spare i found and 6tb hot spare, leaving the read error disks untouched in case something fails.  Sounds like i am still at risk if a 3rd disk fails during rebuild.  I turned off dockers to minimize activity during this time period.  It will take 1.5 days or more for the rebuild (array has 19 disks, 1 cache and 2 parity).