OpenSolaris derived ZFS NAS/ SAN (OmniOS, OpenIndiana, Solaris and napp-it)

_Gea

2[H]4U
Joined
Dec 5, 2010
Messages
3,913
If this is napp-it on Solarish (OmniOS or Oracle Solaris)

If napp-it can see the disks within the VM, you can import the pool. Otherwise the pool is not available.

If this is napp-it on Linux, napp-it should see the disks as then napp-it runs native on Linux (with a reduced feature set compared to the Solaris/Unix version), see https://napp-it.org/downloads/linux_en.html
 

_Gea

2[H]4U
Joined
Dec 5, 2010
Messages
3,913
For a commercial cloud storage Amazon S3 is a de facto standard,

MinIO is an S3 compatible opensource cloud service, ultrafast and scalable nearly without limits and really easy to setup and maintain without the complexity of other cloud services. Basically the whole miniIO S3 server is a single file. Not the complexity of the solutions based on an Apache AMP stack.

As OmniOS has included minIO in the extra repository.
I have added support into napp-it.

You can now share a ZFS filesystem locally via NFS/SMB
and additionally for the Internet via S3 per ZFS filesystem.

Details, see https://forums.servethehome.com/ind...atible-client-server-minio.27524/#post-257208
 

Aesma

[H]ard|Gawd
Joined
Mar 24, 2010
Messages
1,850
Hello all.

I've been running nappit on OI for years.

I've started a replace of the wrong drive by mistake. Usually it wouldn't be a big issue, however I was already replacing another drive (expanding my pool) so now the result is a very slow resilver. The replaces aren't on the same vdev, there is no redundancy issue, I just want to avoid running the server for weeks or more just because of this mistake. Is there a way to cancel the replace ? I can't find it. I tried to offline the new disk, I've now physically removed it, but the "replacing" is still there, and the resilver is still slow as hell (about 100MiB/s instead of the usual 1,2GiB/s).

Code:
NAME                         STATE     READ WRITE CKSUM      CAP            Product /napp-it   IOstat mess
    abc                          DEGRADED     0     0     0
      raidz3-0                   DEGRADED     0     0     0
        c6t5000CCA24CE49B48d0    ONLINE       0     0     0      4 TB           HGST HDN724040AL   S:0 H:0 T:0
        replacing-1              DEGRADED     0     0     0
          c6t5000CCA24CE70C3Fd0  ONLINE       0     0     0      4 TB           HGST HDN724040AL   S:0 H:0 T:0
          c6t5000CCA23DC13FCEd0  UNAVAIL      0     0     0  cannot open                       
        c6t5000CCA24CE78517d0    ONLINE       0     0     0      4 TB           HGST HDN724040AL   S:0 H:0 T:0
        c6t5000CCA24CEA64DBd0    ONLINE       0     0     0      4 TB           HGST HDN724040AL   S:0 H:0 T:0
Thanks for your help
 

_Gea

2[H]4U
Joined
Dec 5, 2010
Messages
3,913
Unlike scrub there is no command to stop a resilver. As a resilver creates a mirror relation you can remove one disk of the mirror unless redundancy is not affected.

btw.
Current Illumos offer sorted/sequential resilver that is much faster than in former releases.
 

Aesma

[H]ard|Gawd
Joined
Mar 24, 2010
Messages
1,850
I can remove or I can't ? It's a raidz3 vdev with the original drive online so there shouldn't be a redundancy problem, right ?

For your other suggestion, would I need to upgrade the pool or just import it in a newer system ? I can't upgrade my OI, it doesn't work (I messed too much with it). My pool is at the latest Solaris compatible version so I'd rather not upgrade it.
 

_Gea

2[H]4U
Joined
Dec 5, 2010
Messages
3,913
In a Z3 you can loose up to three disks without a dataloss

About the running replace
c6t5000CCA24CE70C3Fd0 ONLINE

if you remove this disk, your resilver will stop with state=degraded.
Finally you must replace the faulted disk some time to go back to online state

If you are on ZFS v28 and want to go to current ZFS (v5000 with features):

-Install a new OS (can use a new bootdisk to keep the old one working) either OpenIndiana Hipster 2019.10 (with GUI) or OmniOS 151032 (minimalistic, suited for production use with a stable/lts and regular security fixes, free but optional with a commercial support option)

-Import the pool.

- Update pool to newest (napp-it click on pool version in menu Pools)
 

Aesma

[H]ard|Gawd
Joined
Mar 24, 2010
Messages
1,850
OK I'll remove that disk and start preparing to install Hipster. I use some windows apps with Wine and I need a GUI.

I want to stay on v28 though.

Thanks
 

Aesma

[H]ard|Gawd
Joined
Mar 24, 2010
Messages
1,850
So at first hipster looks like my old OI.

Then I see things are missing.

I'm having much trouble setting up vnc. I installed nappit and saw I could activate vnc with it, good, except where do I setup the password ?

edit : OK I installed again and didn't try to do vnc myself, just used nappit to activate it and it works perfect.
 
Last edited:

Aesma

[H]ard|Gawd
Joined
Mar 24, 2010
Messages
1,850
Now resilvering is set to last about 2 days (probably 3, from experience) so that's good.

I have a strange issue though. My pool had about 1% free space before I changed OS. That is, without reservation, that I had removed, as I need all the space I can get, and performance isn't my main concern. Now, after importing the pool, there is 0 free space, as in not a single bit. All the files seem to be there, but the free space is gone. I deleted a few files, and still, no free space. I vaguely remember having changed some parameter to gain a bit of space, but can't remember what. I don't think it's recordsize, and recordsize shouldn't affect free space anyway, right ?

Code:
NAME  PROPERTY              VALUE                  SOURCE
abc   type                  filesystem             -
abc   creation              Wed May  6 21:47 2015  -
abc   used                  161T                   -
abc   available             0                      -
abc   referenced            486K                   -
abc   compressratio         1.00x                  -
abc   mounted               yes                    -
abc   quota                 none                   default
abc   reservation           none                   default
abc   recordsize            128K                   default
abc   mountpoint            /abc                   local
abc   sharenfs              off                    default
abc   checksum              on                     default
abc   compression           off                    default
abc   atime                 on                     default
abc   devices               on                     default
abc   exec                  on                     default
abc   setuid                on                     default
abc   readonly              off                    default
abc   zoned                 off                    default
abc   snapdir               hidden                 default
abc   aclmode               discard                default
abc   aclinherit            restricted             default
abc   createtxg             1                      -
abc   canmount              on                     default
abc   xattr                 on                     default
abc   copies                1                      default
abc   version               5                      -
abc   utf8only              off                    -
abc   normalization         none                   -
abc   casesensitivity       sensitive              -
abc   vscan                 off                    default
abc   nbmand                off                    default
abc   sharesmb              off                    default
abc   refquota              none                   default
abc   refreservation        none                   default
abc   guid                  14230760374975883861   -
abc   primarycache          all                    default
abc   secondarycache        all                    default
abc   usedbysnapshots       0                      -
abc   usedbydataset         486K                   -
abc   usedbychildren        161T                   -
abc   usedbyrefreservation  0                      -
abc   logbias               latency                default
abc   dedup                 off                    default
abc   mlslabel              none                   default
abc   sync                  standard               default
abc   dnodesize             legacy                 default
abc   refcompressratio      1.00x                  -
abc   written               486K                   -
abc   logicalused           161T                   -
abc   logicalreferenced     16K                    -
abc   filesystem_limit      none                   default
abc   snapshot_limit        none                   default
abc   filesystem_count      none                   default
abc   snapshot_count        none                   default
abc   redundant_metadata    all                    default
abc   special_small_blocks  0                      default
abc   encryption            off                    default
abc   keylocation           none                   default
abc   keyformat             none                   default
abc   pbkdf2iters           0                      default
 
Last edited:

_Gea

2[H]4U
Joined
Dec 5, 2010
Messages
3,913
If there are no snaps, no reservations or zvols, the pool is simply full. Delete some files and associated snaps to regain space. You should enable compress. As it seems you have updated the pool to v5000, you can use the very efficient lz4 compress. This works only for newly created files ex when you move between filesystems that requires a copy.
 
Top