• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

All-in-one-server

G

GreatOcean

n00b
Joined
Sep 11, 2013
Messages
45
Hi everyone,

I'm having difficulties chosing - I've spent too much time reading and getting nowhere.
I have 4x2TB drives aswell as 8x4TB drives and a M1015 IT-flashed card sitting and waiting for me to pull the trigger on an all-in-one install. It will run on a box with 6x Gbit NICS (all Intel and on the HCL list, tested and working). Most file transfers will be sequential in nature.
I've tested the M1015 with passthrough and all works, no issues detected so far, but this was passing through to a Windows 2012 Server OS that I'm familiar with for testing purposes.

FreeNAS, Nas4Free, OmniOS/Napp-IT, Solaris Express 11 with... well you know the alternatives - what gives? I hear much success and also a lot of failures with different combos (stability issues, drive disconnects, etc). I'm not sure which way to go, all looks promising and fits my needs, I just don't wanna get bummed out and have to fight night and day to eventually relocate 30TB of data because of a "bad choice" 1 year from now. Any pointers would be helpful and if possible, the very much appreciated "why" :).

PS. I have thought about running the 4x 2TB drives in Z1 and the 8x 4TB in Z2. Is this a bad line of going with this storage setup?

EDIT1: Sorry for being unclear. I was assuming All-in-One meant ESXI. Host is running ESXI 5.1u1 at the moment and has been for testing, is very stable. Is 5.5 the way to go, as I read so much about hidden features in the webbased cilent and hardware version 10 being inaccessible etc, and that the webbased client means having to pay for vCSA?
EDIT2: I will read up on the vdev-sizes. Apparently I have more reading to do.

Best Regards,
Andreas
 
Last edited:
For All-In-One, the first decision is the hypervisor.
For an All-In-One that supports all Guest operating systems optionally with direct hardware acces,
look at ESXi, especially the new 5.5 free edition without RAM limit.

For the storage part, use ZFS with pass-through enabled for the SAS controller.
For pool-layout prefer ZFS vdevs with 4,8,16,.. datadisks + parity disks
to achieve a maximum of usable capacity. Prefer webbased installations on ESXi
(webmanaged like FreeNAS or ready to use without extra installation or configuration
like my own napp-it ZFS appliance where installation is only a copy of a folder)
 
Does the following still hold true?

FreeNAS’ ZFS stripes data in chunks up to 128 kilobyte . For performance reasons it is preferred that you use these conventions when creating RAIDZ1 and RAIDZ2 VDevs. (n is any whole number you want….1, 2, 3 etc.)
RAIDZ1 should have the total number of drives equal to 2n + 1. (ie 3,5,9, etc. drives for the VDev)
RAIDZ2 should have the total number of drives equal to 2n + 2. (ie 4, 6, 10, etc drives for the VDev)
RAIDZ3 should have the total number of drives equal to 2n + 4. (ie 5, 7, 11, etc drives for the VDev)
>>> This is to ensure ZFS stripes fall on the 4k sector boundary of newer generation hard drives. <<<
If you do not intend to transfer large quantities of data constantly this thumbrule can be disregarded. For instance, if you intend to use your server to stream DVD rips of your collection, this thumbrule can be ignored.
 
GreatOcean said:
Does the following still hold true?

FreeNAS&#8217; ZFS stripes data in chunks up to 128 kilobyte . For performance reasons it is preferred that you use these conventions when creating RAIDZ1 and RAIDZ2 VDevs. (n is any whole number you want&#8230;.1, 2, 3 etc.)
RAIDZ1 should have the total number of drives equal to 2n + 1. (ie 3,5,9, etc. drives for the VDev)
RAIDZ2 should have the total number of drives equal to 2n + 2. (ie 4, 6, 10, etc drives for the VDev)
RAIDZ3 should have the total number of drives equal to 2n + 4. (ie 5, 7, 11, etc drives for the VDev)
>>> This is to ensure ZFS stripes fall on the 4k sector boundary of newer generation hard drives. <<<
If you do not intend to transfer large quantities of data constantly this thumbrule can be disregarded. For instance, if you intend to use your server to stream DVD rips of your collection, this thumbrule can be ignored.
Click to expand...

128k is the default value on any ZFS implementation but you can set others.

there is a good article at
http://www.joyent.com/blog/bruning-questions-zfs-record-size or
https://blogs.oracle.com/roch/entry/tuning_zfs_recordsize

Without any tuning, using 4,8,16,.. datadisks per vdev (+ parity disks) is recommended
mostly because of capacity reasons.
 
Thanks for the links Gea.
I believe I have misunderstood a tremendous amount of things regarding ZFS and how it works. In fact I'm now very confused and need to go back to literature.

I think my main mistake was thinking Raid5 = Z1 and Raid6 = Z2 but I'm fairly sure they're very very far apart as you mention parity disks? I thought parity was spread over all the disks assigned to the vdev?

If we assume just for an example to remove some confusion on my side, I pass through the M1015 IT-flashed card. I hook up 8x 4TB drives.
I want to create a vdev of all the drives, raidz2 pool, would that mean 6 drives for vdev + 2 parity disks, or would that mean 8 drives for vdev and the need of 2 parity disks? or are parity disks optional?
More links are always appreciated :)
 
ZFS Raid-Z spread parity over all disks just like Raid-5/6 but without their write-hole problem.
The 4,8,16 + parity calculation is only to have an easy way to calculate total number of disks per vdev

ex in z3: use 8 datadisks + 3 parity disks=11 disks or in z2 use 8 disks + 2 parity disks = 10 disks
to have a golden number.

These numbers are also not per pool but per vdev.
 
Thanks for clearing a lot up for me Gea,

Could you please provide me with a link or information about these drive numbering golden numbers? I find too much conflicting data and I sure don't understand why performance would be slower and less % usable space if I'm not using the "golden numbers"?

I also find a lot of information just like this:
"Keep pool capacity under 80% for best performance. Due to the copy-on-write nature of ZFS, the filesystem gets heavily fragmented."

Is this really true?

Best regards and thanks for your patience
 
The "golden number rule" is mainly a capacity problem on modern 4k disk.
Google for "raid-z 128k wasted space 4k disk" and you will find all needed infos.

Perforrmance on nearly full disks or pools is a problem with every filesystem.
Copy On Write filesystems where every modified datablock is newly written and not partly updated are more sensitive to this. So yes, this is true.
 
That is a lot of disks.

You better make sure you're talking to the right people. If I were you it'd be zfs, bare metal, mirrored disks
 
Trying to make it all work atm, Downloaded the All-in-One preconfigured from here:
http://www.napp-it.org/downloads/index.html

Installed like a charm.

EDIT: I think I'm in over my head with Napp-it/OmniOS - I get a lot of weird warnings that I think are smart related on new drives.
From what I can tell, it looks like the smart codes sent to drives are SCSI, and not SAS/SATA codes, and they show up, a lot. I can't copy/paste so bear with possible spelling errors:
scsi: [ID 107833 kern.notice] ASC: 0x0 (<vendor unique code 0x0>), ASQ: 0x1d, FRU: 0x0
WARNING: /scsi_vhci/disk@g5000c5006536df47 (sd3):
Error for Command: <undecoded cmd 0xa1> Error Level: Recovered

This for every disk recognized by the system including the systemdisk.

I also can't find a way to navigate to settings, as the web interface just blinks rapidly on certain menus, like "System -> HW and Localization -> Localization", The OS thinks I'm in Africa and I'm not (and I can't set the time).
I can't start any service it seems, I don't know why as I get no error messages, on dropdowns they're shown as disabled, I chose enable, screen blinks, and it's back as disabled.
I'm sure this is an error because of the middleman (me) between chair and keyboard, so it might be a too steep learning curve as I want my drives up and running for testing. Any suggestions of something that's more userfiendly (intuitive) and easy to use (ie RTFM less than 1k pages) :)?
 
Last edited:
Cannot comment about the disk messages
but on localization, you need to choose a region like Europe,
wait until the towns are loaded and select a town.

On services like SMB or NFS, you need to create a pool first.
If you create a filesystem with sharing enabled, the service starts automatically
(You need the service menu only in case or problems)
 
Hello Gea, you're coming to the rescue again :)

Thing is, I can't chose anything, page loads while it says:
Processing, please wait...

Town not valid, reload
Here the page reloads, and reloads, and reloads.
This particular case it's under Home - System - HW and Localization - Localization
There were more tabs behaving just like this one, and the only service started from start is ssh.

DL:ed (I think it is your made package?), now on clean ESXI 5.5, DL'ed it again if in case it was broken. I'm sure I must be doing something wrong. As I'm only looking to try now for me this is not an issue, I just want to find out of my hardware will perform as I want it to, but as this seems to be the "thing" to run (and for me it fails) it makes me a bit worried to bring 40tb raw space online :p

Regarding the drive messages, it seems to be a known problem (this is much people having the issue already from 2011 and forward), where the smartmon package is identifying drives wrongly (as I have understood it). No known solution though (that I have found). But the same behaviour happens on all M1015 flashed with IT-mode for me. I have a BIOS on it from 28th july this year (unfortunately on each card, so cannot isolate), cannot remember FW no w/o rebooting. 17.x-something?

Atm for testing purposes I have tested 2x 300gb SAS drives, 1 SSD and 1 WD Red 4TB, same message on any of these drives, aswell as the virtual image I'm booting from (as that is sd1). I've tried and rotated the other 3 WD Red 4TB, same behaviour and I have another 8 Seagate 4TB NAS disks, only two tested, and same behaviour. I'm clueless and since this is my first faceoff with any solaris-ish based OS, I have no idea where to go next.

I've tested debian 7.1.0 and smartmon is not misbehaving there.

I guess I will have to try freebsd-ish based OS and see if I am more lucky there.

EDIT: Would it be a smart idea to install omnios on bare metal and see if errors reproduce just for isolation?

EDIT2:
Noticed some "things" that I don't know what to make of. I assume (and possibly wrongly?) that hardware version is 9 of the virtual machine package as it runs under 5.1u1.
Under 5.1u1 I have the smart errors - but passthrough of onboard sata works - and the image is way faster (snappier), and the web interface aswell
Under 5.5 I don't have the smart errors (second machine up 2 hours, no smart errors). Passthrough is however broken. Changing back to 5.1u1 = smart error codes within 30 minutes.
So erm... I will test some more before giving up. But the web-interface "bugs" are there both in 5.1u1 and 5.5. Can anyone confirm this? IE10/IE11, Firefox, Chrome and Safari tested. Same behavior (and safari is uberslow in the web-ui so lets just ignore that).
 
Last edited:
known problems with 5.5

e1000 vnic:
You must disable TCP segmentation offload or you get massive performance problems
up to data corruption. Maybe similar to
http://kb.vmware.com/selfservice/mi...nguage=en_US&cmd=displayKC&externalId=2058692
I have disabled TCP segmentation in my downloadable esxi appliance 13b

vmxnet3 seems ok.

pass-through of AHCI not working

known problems with napp-it
- napp-it loads basic disk, pool and partition infos on every request, this can give problems with
- parted (hangs)
mostly due to disks without valid partitions (initialize with menu disks initialize)
- bad disks (no reaction within web-ui)

check parted, zpool status or format at console

smartmontools are ok beside the softerror counter goes up on every check

Localization ex
Set zone to US, wait until page is auto-reloaded with towns/states and check state then ex Alaska.
 
Hello Gea,

The page has now refreshed 13.200 times and counting. 1 refresh per second.
It just sits there, two lines of text. Nothing to chose. Waiting doesn't help :)

"Processing, please wait...

Town not valid, reload"

In the bottom, there is the same text scrolling by, listing every known timezone in /usr/share/lib/zoneinfo, and it claims Africa is wrong. This is also the first listed alphabetically inside that folder, so I suspect something is not right. I've tried several installs, all have the same behaviour.

I will try w/o esxi aswell, just to see what is going on, once I get the time for this.
FreeNAS 9.1.1 is up and running atm for testing purposes, not great performance under 5.5, very fast under 5.1u1 here.
 
GreatOcean said:
Town not valid, reload"
Click to expand...

I rechecked the script.
It worked when using zone/region like "US/Alaska" but gives an error
on settings without a region like CET.

I fixed that in current 0.9b3 and 0.9d1
Goto update and select download to reload the newest release.
 
Thank you, will test it out once I get home :)

EDIT: Worked awesome, thanks again Gea!
Best Regards,
 
Last edited:
OK, did a Bonnie test from Napp-it - I think there's a small bug in result presentation as the webpage says "running" forever :)

Anyway, does this look normal? It is being run with 4vcpu, 12GB ram, on i7 3770, M1015, 8 drives (5900 rpm drives in raidz2 (not golden, 2 more drives are coming, just testing)


NAME SIZE Bonnie Date(y.m.d) File Seq-Wr-Chr %CPU Seq-Write %CPU Seq-Rewr %CPU Seq-Rd-Chr %CPU Seq-Read %CPU Rnd Seeks %CPU Files Seq-Create Rnd-Create

tank 29T start 2013.10.09 24G 395 MB/s 87 604 MB/s 40 303 MB/s 29 279 MB/s 75 538 MB/s 19 932.2/s 1 16 +++++/s 26262/s

EDIT: Maybe a bit better readout from Bonnie.log

root@Napp-it-13b:/tank# cat Bonnie.log
Version 1.03e ------Sequential Output------ --Sequential Input- --Random-
-Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP
2013.10.09 24G 395881 87 604354 40 303585 29 279843 75 538804 19 932.2 1
------Sequential Create------ --------Random Create--------
-Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP
16 +++++ +++ +++++ +++ +++++ +++ 26262 100 +++++ +++ 18485 100
2013.10.09,24G,395881,87,604354,40,303585,29,279843,75,538804,19,932.2,1,16,+++++,+++,+++++,+++,+++++,+++,26262,100,+++++,+++,18485,100

EDIT2:
Memory size: 12288 Megabytes

write 12.8 GB via dd, please wait...
time dd if=/dev/zero of=/tank/dd.tst bs=2048000 count=6250

6250+0 records in
6250+0 records out

real 20.7
user 0.0
sys 2.5

12.8 GB in 20.7s = 618.36 MB/s Write

wait 40 s
read 12.8 GB via dd, please wait...
time dd if=/tank/dd.tst of=/dev/null bs=2048000

6250+0 records in
6250+0 records out

real 16.5
user 0.0
sys 2.6

12.8 GB in 16.5s = 775.76 MB/s Read
 
Last edited:
Okay, so this nub (me) is running into some oddball errors.

SMB to the Omnios/Napp-it box is so much faster than NFS for some reason (magnitude massive), I can't figure out why. SMB maxes out the gbit connection on a steady 112mbyte/s copy operation. NFS, if lucky, peaks at 70-75mbyte/s, then keeps dropping. I get a lot of write-waits (copy operation stops, and just sits there for a few seconds, then it resumes).

If I mount an nfs share on any linux box I have (ubuntu, debian or centos), doing a ls on the share takes about 30s, then any consecutive ls is instant.

I've ironed out
nic's, they're working great.
I've ironed out that I have no speed difference using e1000 or vmxnet3, only cpu usage difference.
raidz pools work great (>600mbyte/s write speed)
cpu usage i'm unsure about, esxi says it uses 20ghz, napp it shows very high cpu usage aswell (too high in my opinion, ~55% on SMB transfers, 35% on NFS transfers, with vmxnet3, +5% on both using e1000)

I haven't found a way (even though it says I should be able to) set rights on shares from a windows box. I have applied credentials when clicking on the share, so it should have accepted me as admin of the box.

For NFS, it keeps complaining that it cannot do chmod on the file/ folder after each copy operation. I'm not sure what to do.
For using SMB shares owner of all files become syslog uucp
For using NFS shares owner of all files become the current linux user:user
But even as admin over SMB, i cannot take ownership or change the owner, or rights, of files/ folders made over NFS
The everyone user has full rights, but for (to me) unobvious reasons, cannot be used by any windows machine (need to supply user/password or is rejected), but any NFS connecting machine can. Which begs the question, how can the everyone user with full rights not chmod, but the file is owned by linux user:group, and the files sent from windows boxes are still owned by system:uucp no matter which user or which windows box?

My user has all rights imaginable at this point in time. Basically, this means that using the NFS share for esxi is simply impossible.

I've struggled with this since my last post and I just can't get it working.
Any pointers?

EDIT: Added some info.
 
Last edited:
_Gea said:
known problems with 5.5

e1000 vnic:
You must disable TCP segmentation offload or you get massive performance problems
up to data corruption. Maybe similar to
http://kb.vmware.com/selfservice/mi...nguage=en_US&cmd=displayKC&externalId=2058692
I have disabled TCP segmentation in my downloadable esxi appliance 13b

vmxnet3 seems ok.

pass-through of AHCI not working

known problems with napp-it
- napp-it loads basic disk, pool and partition infos on every request, this can give problems with
- parted (hangs)
mostly due to disks without valid partitions (initialize with menu disks initialize)
- bad disks (no reaction within web-ui)

check parted, zpool status or format at console

smartmontools are ok beside the softerror counter goes up on every check

Localization ex
Set zone to US, wait until page is auto-reloaded with towns/states and check state then ex Alaska.
Click to expand...

That's good to know, I hadn't seen the e1000 issue. Unfortunately I don't know much about NFS so can't offer much help. I have an NFS target on my QNAP that I point at but I haven't done much testing with it.
 
GreatOcean said:
For NFS, it keeps complaining that it cannot do chmod on the file/ folder after each copy operation. I'm not sure what to do.
For using SMB shares owner of all files become syslog uucp
For using NFS shares owner of all files become the current linux user:user
But even as admin over SMB, i cannot take ownership or change the owner, or rights, of files/ folders made over NFS
The everyone user has full rights, but for (to me) unobvious reasons, cannot be used by any windows machine (need to supply user/password or is rejected), but any NFS connecting machine can. Which begs the question, how can the everyone user with full rights not chmod, but the file is owned by linux user:group, and the files sent from windows boxes are still owned by system:uucp no matter which user or which windows box?

My user has all rights imaginable at this point in time. Basically, this means that using the NFS share for esxi is simply impossible.

I've struggled with this since my last post and I just can't get it working.
Any pointers?

EDIT: Added some info.
Click to expand...

With SMB, you have user authentification.
NFS3 has no user authentification or verification.

If you connect NFS shares and create a file, the owner can be different
on different clients/servers or with different settings sometimes as nobody
or the UID of the client computer - all on a goodwill base.

Result: SMB and NFS are basically incompatible regarding permissions.
NFS is perfect when you need a highspeed network filesystem in a secure
environment ex. ESXi NFS datastore or a secure network with anonymous access.

What you can do:
- connect SMB as root to modify permissions (root is owner not a Windows admin)
ans set ACL/ permissions recursivly to everyone@=modify
- on OmniOS, set aclmode to discard. This will hinder NFS to modify permissions
and use ACL from a parent folder with inheritance
 
Ah ok, that clears some of the issues up and thanks for the input.

Any input on the CPU usage (is it normal?)/ slow NFS?

I did a bunch of small file copying aswell. Here (folder of ~10k files/ 4gb) SMB completes the transfer about 100 times faster. After each file, NFS copy pauses, then continues. Never seen this behavior and don't know what to look for or tweak.

So basically SMB speed is very OK, NFS not so much. Tried tweaking rsize and wsize but not helping much if at all.
 
GreatOcean said:
Ah ok, that clears some of the issues up and thanks for the input.

Any input on the CPU usage (is it normal?)/ slow NFS?

I did a bunch of small file copying aswell. Here (folder of ~10k files/ 4gb) SMB completes the transfer about 100 times faster. After each file, NFS copy pauses, then continues. Never seen this behavior and don't know what to look for or tweak.

So basically SMB speed is very OK, NFS not so much. Tried tweaking rsize and wsize but not helping much if at all.
Click to expand...

CPU usage - only ESXi is relevant.
NFS write weakness - set sync property of the filesystem to disabled and try again
 
Any idea what would be normal?
OmniOS/Napp-it right in esxi 5.1u1, vmxnet3 adapter, 16gb ram, testing array of 6x 4tb drives in raidz2
One user copying files to the server @ 80mbyte/s over NFS
OmniOS is given 4 out of 8 cores of a FX8350, and 16gb ram

CPU usage in esxi is a steady 11ghz. I copy two files to it, it flats out on 20ghz (all cores max used)
 
Anyone know about expected CPU usage?
I can't seem to find a way to lower it. I know it's supposed to be CPU intensive but I don't even have deduplication going... on writing from single or two clients (big files @ 20gb each), ESXI eats all CPU given. Odd?

Also, I've had the test machine up and running for some time now, continuously (no reboot)
After filling the array to 60%, Read performance has dropped about 30% (can't go over 500mb/s now) and write performance is lower than 200mb/s. SMB performance also lowered a lot, from always staying within or above >100mbyte/s for large sequential transfers, it's about 70mbyte/s now. NFS is still enabled, but I stopped using it.

I think this must be unexpected behavior, but I can't find anything wrong at all. Where do I start looking?
 
_Gea said:
What you can do:
- connect SMB as root to modify permissions (root is owner not a Windows admin)
ans set ACL/ permissions recursivly to everyone@=modify
- on OmniOS, set aclmode to discard. This will hinder NFS to modify permissions
and use ACL from a parent folder with inheritance
Click to expand...

You also mention "NFS write weakness - set sync property of the filesystem to disabled and try again", any pointers to where I would make this input, aswell as the aclmode to discard?

Oracle/ Solaris references aren't much of use and those referring to linux/ BSD generally are very explanatory but as commands (or syntax) are not alike, linux/ BSD references "generally" cannot be used.

PS; I decided to move away from esxi/ all-in-one, all other performance related "issues" are now gone. :) Reads/Writes are about 1gbyte/ second in both dd/bonnie++, I can easily saturate 6x gbit nics for very long times without hickups or performance issues (I could not when virtualized, performande degratation each day, every day). I expect my hardware wasn't up to par doing this all virtualized, but it all sorted itself out.
 
Thanks danswartz, is it too trivial of me to assume that dataset is a zfs filesystem? I have a hard time making sense of the terminologies, further adding to my insecureness; as:

root@knox:/tank/data/esxi# zfs
missing command
usage: zfs command args ...
where 'command' is one of the following:

create [-p] [-o property=value] ... <filesystem>
create [-ps] [-b blocksize] [-o property=value] ... -V <size> <volume>
destroy [-fnpRrv] <filesystem|volume>
destroy [-dnpRrv] <filesystem|volume>@<snap>[%<snap>][,...]

snapshot [-r] [-o property=value] ... <filesystem@snapname|volume@snapname> ...
rollback [-rRf] <snapshot>
clone [-p] [-o property=value] ... <snapshot> <filesystem|volume>
promote <clone-filesystem>
rename [-f] <filesystem|volume|snapshot> <filesystem|volume|snapshot>
rename [-f] -p <filesystem|volume> <filesystem|volume>
rename -r <snapshot> <snapshot>
list [-rH][-d max] [-o property[,...]] [-t type[,...]] [-s property] ...
[-S property] ... [filesystem|volume|snapshot] ...

set <property=value> <filesystem|volume|snapshot> ...
get [-rHp] [-d max] [-o "all" | field[,...]] [-t type[,...]] [-s source[,...]]
<"all" | property[,...]> [filesystem|volume|snapshot] ...
inherit [-rS] <property> <filesystem|volume|snapshot> ...
upgrade [-v]
upgrade [-r] [-V version] <-a | filesystem ...>
userspace [-Hinp] [-o field[,...]] [-s field] ...
[-S field] ... [-t type[,...]] <filesystem|snapshot>
groupspace [-Hinp] [-o field[,...]] [-s field] ...
[-S field] ... [-t type[,...]] <filesystem|snapshot>

mount
mount [-vO] [-o opts] <-a | filesystem>
unmount [-f] <-a | filesystem|mountpoint>
share <-a | filesystem>
unshare <-a | filesystem|mountpoint>

send [-DnPpRv] [-[iI] snapshot] <snapshot>
receive [-vnFu] <filesystem|volume|snapshot>
receive [-vnFu] [-d | -e] <filesystem>

allow <filesystem|volume>
allow [-ldug] <"everyone"|user|group>[,...] <perm|@setname>[,...]
<filesystem|volume>
allow [-ld] -e <perm|@setname>[,...] <filesystem|volume>
allow -c <perm|@setname>[,...] <filesystem|volume>
allow -s @setname <perm|@setname>[,...] <filesystem|volume>

unallow [-rldug] <"everyone"|user|group>[,...]
[<perm|@setname>[,...]] <filesystem|volume>
unallow [-rld] -e [<perm|@setname>[,...]] <filesystem|volume>
unallow [-r] -c [<perm|@setname>[,...]] <filesystem|volume>
unallow [-r] -s @setname [<perm|@setname>[,...]] <filesystem|volume>

hold [-r] <tag> <snapshot> ...
holds [-r] <snapshot> ...
release [-r] <tag> <snapshot> ...
diff [-FHt] <snapshot> [snapshot|filesystem]

Each dataset is of the form: pool/[dataset/]*dataset[@name]

For the property list, run: zfs set|get

For the delegated permission list, run: zfs allow|unallow

So for the newly invited to solaris, what a dataset is referenced to isn't entirely clear.
So in my case, where i have
/tank (pool)
/tank/vm

and vm contains esxi vms, I would do "zfs set sync=disabled /tank/vm" ?

Also regarding the "set aclmode to discard", would that mean
"zfs set aclmode=discard /tank/vm" ?

If so, it starts to make a whole lot of sense :)

Thanks for your patience.
 
No leading '/'. So, 'zfs set sync=disabled tank/vm'. Yes, dataset is basically filesystem.
 
You must log in or register to reply here.
Back
Top