TrueNas ISCSI devices showing 0 capacity to vmware

[dacheat]

I recently moved from FreeNas to TrueNas and went to build a new pool to pass to my vmware cluster but something about the new pool isn't working as expected. I have 3 other ISCSI mounts from the same TrueNas Server that have no problems.

I created a new Pool/zvol: https://i.imgur.com/1roJscH.png
I added it to my existing portal/target/extents to match my 3 other ISCSI volumes
The device shows up in the ISCSI vmhba after a refresh, but shows 0 capacity and I have no idea why: https://i.imgur.com/eXzma3q.png

Wondering if someone has any brilliant ideas for this.
Thanks

 

[HoneyBadger]

Can you post your iSCSI target and detailed ZVOL information? Embeds are preferable to links, because imgur mutilates images on mobile (webp compression to 640 wide)

My offhand guess is to check the "hide physical block size" option.

 

[dacheat]

Thanks for your help.
Well I've run into a new issue where any time I try and edit a zvol to look at the settings it always opens a different zvol...

Datastore settings:

Extent:

target:

 

[HoneyBadger]

Thanks for your help.


Extent: View attachment 47608

Check off "Disable Physical Block Size Reporting" - save, rescan the HBA from VMware. See if that doesn't sort it for you.

 

[dacheat]

So I did some more testing.
Passed a non-pool/non-zvol via iscsi. Worked as expected
I connected to the truenas from a windows computer and was able to mount the cached drive I'm having trouble with but could not initialize it, showed an unknown amount of storage.
I demolished the pool and recreated it with the same mirrored cache and ran into the same problem.
I demolished the pool and recreated it without the mirrored cache, just 3 disks and didn't have any problems.

 

[dacheat]

Doesn't seem like I can edit my previous post.
I demolished the pool and recreated it with a single striped cached drive, ignoring the warnings about data loss and that seemed to work as expected as well. Seems like there's actually a problem with one of the drives I'm passing into the mirrored write cache, if I try and mount that as it's own iscsi target it has the same 0 space problem.

 

[HoneyBadger]

A bad cache/log device shouldn't be impacting a ZVOL in that manner unless there's been a power outage.

Side note - your parent dataset is using sync=standard - for iSCSI this isn't sufficient to guarantee safe writes. You'd have to override that to sync=always for the dataset or each ZVOL, and you should expect a performance impact, potentially a significant one depending on the devices you've got in use.

Can you post the full system details including hardware specs, and the output of zpool status -v?