TrueNAS Scale - SLOG Over-Provision

[Michael Aos]

I'm planning to build a physical TrueNAS Scale box, but in the meantime I've been playing with it on VMware.

I've set up 2x 480GB disks for SLOG Over-Provision. Also looked found additional info on resize-command on Documentation Hub. Thinking I would resize each to 16GB and mirror them.

I've been tinkering with this for a while and eventually figured out how to remove the SLOG from the active pool. Then I figured out how to offline the drives.

• diskname1 and diskname2 are the names of the disk you want to resize.
• number is the size (in gigabytes) to resize the disk(s) to.
• true/false in the sync property enables or disables synchronizing the new size of the disk(s) with the database cache.
• true/false in the raise_error property sets whether the disk(s) raise a CallError upon failure (true) or only log errors (false).

I figured out how to get to System Settings:Shell, then type 'cli'.
admin@truenas[~]$ cli
[truenas]>storage disk resize disks={"name":"sdd,sde", "size":"16"} sync=true raise_error=true
[0%] ...
Error: Failure resizing: DISK: 'sdd,sde' SIZE: 16 gigabytes ERROR: Command '('disk_resize', 'sdd,sde', '16G')' returned non-zero exit status 1.

What the heck am I doing wrong?

[truenas]> storage disk retaste
[0%] ...
[85%] Retasting disks...
[95%] Waiting for disk events to settle...
[100%] Retasting disks done...
SUCCESS

[truenas] storage disk> query
+--------------------------------------------+------+-----------+--------+--------+--------+----------------+-------------+--------------+------------+--------------+-------------+--------------+------------+----------+------------+---------------+--------------+--------------+------+----------------------+------+---------+-----------+----------------+--------+
| identifier | name | subsystem | number | serial | lunid | size | description | transfermode | hddstandby | advpowermgmt | togglesmart | smartoptions | expiretime | critical | difference | informational | model | rotationrate | type | zfs_guid | bus | devname | enclosure | supports_smart | pool |
+--------------------------------------------+------+-----------+--------+--------+--------+----------------+-------------+--------------+------------+--------------+-------------+--------------+------------+----------+------------+---------------+--------------+--------------+------+----------------------+------+---------+-----------+----------------+--------+
| {uuid}2491dd10-cc6a-4751-8592-e696baf2e806 | sda | scsi | 2048 | | <null> | 21474836480 | | Auto | ALWAYS ON | DISABLED | true | | <null> | <null> | <null> | <null> | Virtual_disk | <null> | SSD | <null> | SCSI | sda | <null> | <null> | <null> |
| {devicename}sde | sde | scsi | 2112 | | <null> | 515396075520 | | Auto | ALWAYS ON | DISABLED | true | | <null> | <null> | <null> | <null> | Virtual_disk | <null> | SSD | <null> | SCSI | sde | <null> | <null> | <null> |
| {devicename}sdd | sdd | scsi | 2096 | | <null> | 515396075520 | | Auto | ALWAYS ON | DISABLED | true | | <null> | <null> | <null> | <null> | Virtual_disk | <null> | SSD | <null> | SCSI | sdd | <null> | <null> | <null> |
| {uuid}373478ba-8e96-4437-8fa6-3a13f18732c2 | sdc | scsi | 2080 | | <null> | 22001227671552 | | Auto | ALWAYS ON | DISABLED | true | | <null> | <null> | <null> | <null> | Virtual_disk | <null> | SSD | 13826010829014269233 | SCSI | sdc | <null> | <null> | <null> |
| {uuid}6745e816-f80f-48dd-87ff-f576aa0cf99b | sdb | scsi | 2064 | | <null> | 22001227671552 | | Auto | ALWAYS ON | DISABLED | true | | <null> | <null> | <null> | <null> | Virtual_disk | <null> | SSD | 3263109309399002586 | SCSI | sdb | <null> | <null> | <null> |
+--------------------------------------------+------+-----------+--------+--------+--------+----------------+-------------+--------------+------------+--------------+-------------+--------------+------------+----------+------------+---------------+--------------+--------------+------+----------------------+------+---------+-----------+----------------+--------+

 

[HoneyBadger]

Hey @Michael Aos

The disk_resize uses the SATA/SAS commands for the host protected area/dynamic disk resizing. This isn't supported by a VMware virtual disk (as shown in your query output) so the call fails. You can add the full-sized SLOG to it for your testing purposes in VMware right now.

In the meantime, have you read through the "Yes, You Can (Still) Virtualize TrueNAS" blog? This goes into a bit of detail about things like PCI hardware passthrough (which would let you directly address the disks.)

 

[Etorix]

480 GB looks like consumer Optane (900p/905p?), which you cannot overprovision.
Optane DC P4800X/P4801X/P5800X can be overprovisioned with Intel utility or standard NVMe tools (nvmecontrol on CORE).

 

[Michael Aos]

Thank you both for the replies.

I was highly suspicious the error was related to the Virtual Disks. I just hadn't been able to corroborate that anywhere.
I was only using the 480GB size / number to approximate a pre-built configuration like the iXsystems TrueNAS Mini X or XL+.

I've got a couple old ML310e Gen8's (Intel(R) Xeon(R) CPU E3-1220 V2 @ 3.10GHz) 32GB RAM laying around to play with while I decide how far I want to take it.

I bought a pair of 22TB WD Red Pro NAS Hard Drives (on sale for $669.98/pr) for starters.

Any specific suggestions on a specific SLOG device if I were to purchase one specifically for this? I have a few Samsung 850 Pro 512GB SATA and OWC Mercury EXTREME Pro drives laying around - they may not even be worth messing with?

 

[HoneyBadger]

Optane drives as suggested by @Etorix are probably among the best consumer-available options. Other options do exist, but generally "enterprise write-intensive, with power-loss-prevention" is what you're looking for.

SLOG benchmarking and finding the best SLOG

I'd like to take a few minutes to talk about SLOG devices and what makes good ones versus bad ones. I have no doubt that this will be a controversial topic since this is not well understood by many people. In short, there's 3 things that you need for a "great" SLOG: 1. Fast throughput 2...

www.truenas.com

But before digging too deep, ensure that you actually do need the SLOG device - you only require one for doing synchronous writes, such as a remote database or VM hypervisor host, where your remote writes "can't be replayed"

 

[Michael Aos]

But before digging too deep, ensure that you actually do need the SLOG device - you only require one for doing synchronous writes, such as a remote database or VM hypervisor host, where your remote writes "can't be replayed"

Thank you!

I've read this a few times, but it's just now starting to click.

By default, ZIL does not handle asynchronous writes. System memory handles these like any standard caching method. This means the ZIL only works for select use cases, like database storage or virtualization over Network File System (NFS).