SLOG Device and zfs tuning in scale

[Kailee71]

Hi all,

after having used Core for quite a few years I am now investigating moving over to Scale. I have been using two 32Gb Optane drives as SLOG to speed up my (sync) writes, and have tuned zfs to allow more dirty data and stretch the txg timeout. Under FreeBSD this was done with kernel tunes;

Code:

truenas# sysctl -a | grep zfs.dirty
vfs.zfs.dirty_data_sync_percent: 20
vfs.zfs.dirty_data_max_max: 4294967296
vfs.zfs.dirty_data_max: 4294967296
vfs.zfs.dirty_data_max_max_percent: 25
vfs.zfs.dirty_data_max_percent: 10

I had a quick look at sysctl -a but it seems no zfs sysctl tuning is present. Can zfs tuning be done in scale, and if so, through the GUI?

Any hints highly welcome.

Kai.

 

[HoneyBadger]

Hey @Kailee71

ZFS tuning can be done in SCALE, through the module parameters under /sys/module/zfs/parameters/ - we're still refining the middleware and UI methods, so for now you'll have to echo VALUE >> /sys/module/zfs/parameters/zfs_foo_bar as a pre/post-init script.

The list of module parameters under the write-throttle tag at the OpenZFS docs is a good place to find their new names, but generally it's "replace vfs.zfs. with zfs_"

 

[Kailee71]

Thank you so much for your speedy response. Really appreciate it!

Will go ahead and try this and report back.

K.

 

[Kailee71]

Sorry it's taking a little longer with the feedback - unfortunately Scale decided it was time for a scrub... Don't want to cancel it as I'll have to go throug it at some stage. As soon as it's done I'll do some (very basic) benchmarking.