Re: Failed adadm RAID array after aborted Grown operation

[Reindl Harald <h.reindl@thelounge.net>]

Am 22.05.22 um 06:13 schrieb Bob Brand:
> Is xfs_repair an option? 

unlikely in case the underlying device has all sort of damages - think 
of the RAID like a single disk and how the filesystem reacts if you 
shoot holes in it

> And, if it is, do I run it on md125 or the
> individual sd devices?

when you think two seconds it's obvious - is the filesystem on top of 
single disks or on top of the whole RAID - the filesystem don't even 
know about individual devices

> Unfortunately, restore from back up isn't an option - after all to where do
> you back up 200TB of data? 

on a second machine in a different building - the inital sync is done 
local and then no matter how large the data rsync is enough - the daily 
delta don't vary just because the whole dataset is huge

and don't get me wrong but who starts a reshape on a 200 TB storage when 
he knows that there is no backup?

> This storage was originally set up with the
> understanding that it wasn't backed up and so no valuable data was supposed
> to have been stored on it. 

well, then i won't store it at all

> Unfortunately, people being what they are,
> valuable data has been stored there and I'm the mug now trying to get it
> back - it's a system that I've inherited.