Available space decreasing but not on replicated drive

markgca

Dabbler
Joined
Nov 7, 2019
Messages
46
i have two identical pools, one for data, one for replication

initially i turned on the recycle function, but then found out i didnt need it due to snapshots; turned it off, and deleted all the recycle bins (i think, maybe there are remnants i missed due to misunderstanding how this works)

have a ton of snapshots, doesnt seem out of line with what i see recommended (5 minutes kept for a few days, weekly kept for months)

but now the available space is down to 3TB, 91% used, but only 25% used of the replicated pool.

i would have thought all those snapshots would be replicated as well, but im new to freenas and dont really understand everything as well as i should.

any thoughts on where to start looking for the issue? i guarantee im not using that much space for data; the replicated one seems about right at 25%. there are a couple extra things im backing up to the main pool, but not another 32TB

thanks for any guidance on this. at 91% the array is stalling now and then and isnt happy. i do know that freenas wants more freespace to work correctly

mark
 

sretalla

Powered by Neutrality
Moderator
Joined
Jan 1, 2016
Messages
9,702
Which of the snapshot jobs is the replication working with?

If you're taking a snapshot every 5 minutes, does that mean you have enough file changes happening to make that sensible? which may mean that the weekly snapshots refer to completely different sets of data to the 5 minute ones?
 

markgca

Dabbler
Joined
Nov 7, 2019
Messages
46
not sure how to answer to be honest
my needs; there are only a couple people accessing this thing at a time, but...the files we work on are important, and losing a day of edits etc would be catastrophic. i had read that snapshots dont take up much space if you dont have many changed files, so i figured i could get away with that.

so not many file changes, but important ones.

is there an easy way to see how much space the snapshots are using? i see the snapshots list, and there are 3800 of them; maybe thats too many? but i dont find anywhere where it shows the space they take up

my first freenas system, so im making lots of mistakes

thanks for your help

edit; decided to remove some snapshots from the snapshot page, just to see what effect they would have. got an error message "items delete failed tank/xyz/xxxx not found"
 

sretalla

Powered by Neutrality
Moderator
Joined
Jan 1, 2016
Messages
9,702
zfs list -t snapshot

The USED column shows how much space it's costing you to have that snapshot... the REFER column is the total size of the dataset/pool referenced at the time the snapshot was taken.
 

markgca

Dabbler
Joined
Nov 7, 2019
Messages
46
so i ran that, and its using up a few TB of data, but not the 30 or so missing
and when i tried to erase them; some would, some gave me a error that it was not found. any ideas on how to fix that? i dont have an issue erasing all of them and starting over if that would help. could the .recycle function which i had enabled before be adding to this? i did go through and delete every .recycle bin i could find
 

sretalla

Powered by Neutrality
Moderator
Joined
Jan 1, 2016
Messages
9,702
Maybe this will help:

At very least you can see quickly where the space has gone.
 

markgca

Dabbler
Joined
Nov 7, 2019
Messages
46
ok, so i just went ahead and erased all the snapshots. many said i couldnt, but if i grouped a few dozen at a time they deleted properly. took a while, but now the available space is where it should be. thanks for all the help. i will redesign my snapshots and at least keep an eye on them

the nice thing about freenas, which i think is really just zfs; i have never been concerned about the underlying data like i was on some prior nas platforms. these types of issues are all nits in the big scheme of things, and it feels like the data is totally secure. it was a good move for us, even though my learning curve is steep and i still cant figure out some basic permissions things, but ill get there. thanks again
 

sretalla

Powered by Neutrality
Moderator
Joined
Jan 1, 2016
Messages
9,702
Also convenient would be to use the % sign to separate the first and last snapshots of a given dataset when using zpool destroy, like this:

zpool destroy pool/dataset@snap001%snap100

That destroys all snapshots in-between, inclusive of the ones identified (uses the snapshot timestamps to determine what's between, so naming is not important as long as you hit the first and last names correctly).

Clearly be very careful when using that command which has the potential to delete a lot of snapshots/data if used incorrectly.
 
Top