Running 6.10.3 here.  I have an old 1TB SSD that, surprise surprise, has started failing.  I tried rsyncing all the data off it, and did a ddrescue as well, but in both cases I got a number of read errors.  So while I have most of the data still, some of it is corrupted.

For now I have switched off Docker and VMs in order to not stress it any further.
I would like to ultimately replace the cache with a mirrored pool of two 1TB SSDs.  I believe (correct me if I'm wrong) that it will be easiest to replace the failed drive with a single new one first, and after that's working, install the additional drive and add it to the pool.

So how would I go about doing that?  I found some instructions for Replacing or upgrading your Cache drive but that seems to assume the original drive is error-free.  A link on Reddit gives instructions that seem straightforward enough, but once again they don't seem to account for the fact that I am replacing a drive with some file corruption on it.

 

So how would you proceed?  Write the ddrescue image to a new drive, and then install that as the new cache?  The problem is, there will be some files that are corrupted on it - I don't care about downloads, but if it's a Telegraf database, then that might be more of an issue.

I found a utility called ddru_findbad that parses the ddrescue log and tells you which files were affected by the read errors - this might give me an idea of what to look out for.

Note I do have "Appdata Backup" installed so could roll back from a previous version if any of my Dockers/VMs are FUBARed.

 

Thanks for any suggestions!

Copy what you can from the old SSD to the new one, any files with read errors won't be copied, so you'll know which files need to be replaced/restored, then add a member to the new pool to make it a mirror.