don't ignore your drives-and dont get them all at once (mercury in retrograde)

[labtopia]

a small cautionary tale to not source all your drives and install at the same time if you can avoid it.

i'm now slowly backing up & resilvering after many attempts a dataset that is about 80tb. i built a storage pod using the backblaze framework of open hardware about 6 yrs ago.

system has had about 6 drives go out in about 6 years but now almost all of the first set of 9 drives are blowing up with smart errors and worse, disconnects...

i've had some existing smart pending errors on 2 drives i've ignored and wish i didnt...dont ignore your drive errors...

glad to still see drives and data and hopefully the resilver replacements keep going ok. at about 30hrs/resilver my fingers are crossed and i'll successfully make it through mercury's retrograde (when it all blew upon april28). Deep breaths...

 

[jgreco]

Sorry to hear it, I am an advocate of heterogeneous pools, primarily for that very reason. Also excessive redundancy.

Good luck, hope it works out.

 

[Mr_N]

What drives did you get all those yrs ago?

 

[labtopia]

howdy! jgreco and mr_n. yeah i have multiple raidz2 that make up the one big ass pool- probably not the best idea, been ok until now but backups and resilvering is going as good as i expect, main important data is backed off- this thing stores almost 20yrs of our post production work and history.

for the drives they are various versions of the seagate 2tb 7200 eco drives. had several go out within their 3yr warranty, but all are now way out of warranty. good thing now is they are around 60bucks to replace, also they spin 2/4/7 so none of this is a surprise but only inconvenience i hope. i'll be more active when my daily sends me a message about drive pending errors

issue i had is freenas fails to boot if it sees a completely failed drive that's hangin up one of the port multipliers, then making the port node unhappy.

fingers crossed, but it's my birthday so im not worrying about it any more today and going to find some fine seafood,

dave

 

[jgreco]

Yeah, that's one of several reasons not to use port multipliers. They're crap grade hardware, which isn't really what you want for reliable storage.

 

[labtopia]

YEP, but built on the same components of backblaze. been ok for so long, i think i let these smart errors pile up which then get exacerbated by the drives banging as they resilver. the plane is setup with 12 or so multipliers going into 3 controllers. I also i found the landlord wirelessly changed the air cooling and some components were compromised by hot temp. Think server with 45 drives running in 80 ambient...

the port resets and not recovering definitely one of the probs.

 

[zambanini]

using eco/green drives is not an issue coming from the stars (mercury retrograde...) - just bad architecture.

 

[labtopia]

ha funny and true- 3 drives replaced and 1 more to resilver- pool online still with data... will keep it up whilst backing more up...