This is something that I have been troubleshooting for some time now and have been searching forum for possible answers

 

When transferring large files to unRaid(1), upload to server starts and then stalls for a few minutes.  Then eventually, says an error has occurred, but when I select SKIP, the file has successfully transferred to unRaid1.

 

 

I have a second unRaid(2) server (on same network/switch) and file transfers to it work fine. 

Howerver, transferring from unRaid2 to unRaid1 again displays same behavior.

 

On unRaid1 I have tried different NIC (am now using both bonded), changed out SATA cables, tried connecting drives to both motherboard SATA data and PCI SATA (HBA)

 

Have also tried switching Boot USB with an new USB drrive

All w/o success

 

Both unRaids are using SSD ZFS (mirror) cache - unRaid1 has 50GB out of 1TB

 

unRaid1-diagnostics-20240822-1216.zip

There's nothing relevant logged, suggesting a possible external issue, same thing happens transferring to the array or pool?

OMG - just transferred a 3GB file to two different drives in array and worked w/o problem.

So, the problem is with the SDDs that I am using?

Could be, you also try to destroy that pool and then create two single device pools and retest.

Just to make sure I am understanding -

Move data from present cache - destroy and recreate cache as a second pool (mirrored) or create pool2 and pool3 separate pools with the SSDs

I also do have a couple SSDs on hand (differednt brand) that I could replace with

40 minutes ago, uglytuna said:

or create pool2 and pool3 separate pools with the SSDs

This, for testing.

 

24 minutes ago, uglytuna said:

I also do have a couple SSDs on hand (differednt brand) that I could replace with

You can also try this first.

Just completed switching out SSDs (from Crucial to WD) and resilvering completed

SMB transfer of 3GB completed w/o problem

Docker update of immich completed from GUI w/o problem

 

Would have never guessed that one of the SSDs was bad.  They weren't that old and extended SMART test didn't reveal any real problems

 

Thanks again

40 minutes ago, uglytuna said:

Would have never guessed that one of the SSDs was bad. 

Possibly not bad, it may be ok after a full blkdiscard.

Am running Spinrite on both drives now.

Will blkdiscard do something better/different?

 

It's the command that tells the drive's firmware to pretty much start over with data allocation, it can help if there are cells that couldn't be optimized for some reason or another.