SMB issues the past few months.

[diskdiddler]

Hi,

I've had some real odd issues with SMB the past few months, generally writes.
I've seen many a post of this on the internet and ignored them because it hasn't majorly impacted me, but overall, it does seem to simply be worse.

I have a variety of machines which talk to TrueNAS over SMB (a lot) and they've started misbehaving. I saw people claim it's ever since 13 but I don't know for certain.
qbittorrent, syncthing, etc, a lot of things trying to work with SMB absoloutely have issues.

My little docker machine has 4 cores, 16GB, Intel 6xxx processor, it does pretty lite stuff but it's absoloutely dying, trying to write to TrueNAS.
Is there any optomisations I can investigate? - I just saw 40% CPU time on TrueNAS with syncthing trying to write files over SMB and other docker containers, on the syncthing server becoming unusable.

Is there a discussion on this I can tune in to?

 

[Samuel Tai]

How full is your pool?

 

[diskdiddler]

How full is your pool?

It's 79% - but this is particularly bad.

 

[Samuel Tai]

The path to success for block storage

It seems like I haven't written a sticky for awhile, but just in the last week I've had to cover this topic several times. ZFS does two different things very well. One is storage of large sequentially-written files, such as archives, logs, or data files, where the file does not have the middle...

www.truenas.com

You’ve only just noticed, but this is almost entirely due to ZFS fragmentation. According to the link, write performance starts going south at 50% occupancy. You’ll need to expand your pool with larger drives to get below 50%.

 

[diskdiddler]

The path to success for block storage

It seems like I haven't written a sticky for awhile, but just in the last week I've had to cover this topic several times. ZFS does two different things very well. One is storage of large sequentially-written files, such as archives, logs, or data files, where the file does not have the middle...

www.truenas.com

You’ve only just noticed, but this is almost entirely due to ZFS fragmentation. According to the link, write performance starts going south at 50% occupancy. You’ll need to expand your pool with larger drives to get below 50%.

I have expanded my pool from 6x8 to 6x16TB disks and SMB performance is still atrociously bad.
The whole system behaves 'weird'

Ubuntu VM with multiple docker containers which interact with TrueNAS is sluggish as hell working with SMB. In top I can see SMBD consuming anywhere from 15 to 40% processor time, with 8 cores, was never like this before.


NOTE: I also have a 500GB SSD in the machine which is used to 'scratch' data and it's performing slower than the main array.

 

[diskdiddler]

Alternatively if you could recommend a non destructive benchmark I could perform on my pools, from within a shell, I'd be curious to identify what is going on with my system.

 

[Samuel Tai]

fio is bundled with TrueNAS. However, can you show zfs get sync <name of your pool>? Also, is your data pool set to be your system dataset pool?

 

[diskdiddler]

fio is bundled with TrueNAS. However, can you show zfs get sync <name of your pool>? Also, is your data pool set to be your system dataset pool?

Thanks for the reply, I will do some testing, initially it looks like my SSD performance has absoloutely tanked, I'm going to mess around, reboot, play and benchmark (luckily I learnt some FIO last week for my disk replacements)