FreeNAS 8.3.1-RELEASE

[jpaetzel]

The FreeNAS development team is pleased to announce the immediate
availability of FreeNAS 8.3.1-RELEASE.

Images and plugins can be downloaded from the following site:

http://sourceforge.net/projects/freenas/files/FreeNAS-8.3.1/RELEASE/

FreeNAS 8.3.1 is based on FreeBSD 8.3 with version 28 of the ZFS
filesystem, and features volume based encryption for ZFS.

There have been no major changes between 8.3.1-RC1 and RELEASE, mostly
bugfixes and minor usability improvements to the GUI. See the
release notes for a complete list:

http://sourceforge.net/projects/freenas/files/FreeNAS-8.3.1/RELEASE/README/download

Please familiarize yourself extensively with the encryption features of
FreeNAS before using them. Doing the wrong thing can end up in a state
where the volume is hidden behind very difficult to break AES 256 encryption.

http://doc.freenas.org/index.php/Volumes#ZFS_Encryption

Many modern CPUs feature hardware support for encryption. If hardware support
is available FreeNAS will use it. In these cases the overhead of encryption will
be negligible. For systems without hardware encryption acceleration the
performance impact will vary based on the number of disks being used in the
encrypted volume.

 

[hervon]

Thanks! I'll look into it this evening.

 

[Nik]

Hi there,

just upgraded my FreeNAS from 8.3.0-p1 to 8.3.1. Now my pool is down with the error message (yellow alert) "WARNING: The volume tank (ZFS) status is UNKNOWN".

In View Volumes it say about my pool "Error getting available space Error getting total space STATUS: Unkown".

Anything I can do here?

Some details of my system:
My Freenas is running as a VM on VMWARE ESXi 5.1.
My pool consists of 6 RAW-LUNs with 2GB each on a RAID-Z2. I have a LOG SSD and and a L2ARC SSD both RAW-LUNs.
The VM has 16GB RAM allocated.

zpool import says the following:

Code:

[root@nas ~]# zpool import                                                      
   pool: tank                                                                   
     id: 2599985722204748768                                                    
  state: UNAVAIL                                                                
 status: The pool was last accessed by another system.                          
 action: The pool cannot be imported due to damaged devices or data.            
   see: http://www.sun.com/msg/ZFS-8000-EY                                      
 config:                                                                        
                                                                                
        tank                                            UNAVAIL  missing device 
          raidz2-0                                      ONLINE                  
            gptid/33c13c4e-49ae-11e2-9a7f-000c294ced7d  ONLINE                  
            gptid/3439edca-49ae-11e2-9a7f-000c294ced7d  ONLINE                  
            gptid/34a6920f-49ae-11e2-9a7f-000c294ced7d  ONLINE                  
            gptid/3512b5e0-49ae-11e2-9a7f-000c294ced7d  ONLINE                  
            gptid/3584f095-49ae-11e2-9a7f-000c294ced7d  ONLINE                  
            gptid/3603142b-49ae-11e2-9a7f-000c294ced7d  ONLINE                  
        cache                                                                   
          dsk/da7p2                                                             
        logs                                                                    
          da7                                           ONLINE        

Let me know if you need more info.

I hope you can help me.

Regards,
Nik

 

[William Grzybowski]

Hi there,

just upgraded my FreeNAS from 8.3.0-p1 to 8.3.1. Now my pool is down with the error message (yellow alert) "WARNING: The volume tank (ZFS) status is UNKNOWN".

In View Volumes it say about my pool "Error getting available space Error getting total space STATUS: Unkown".

Anything I can do here?

Some details of my system:
My Freenas is running as a VM on VMWARE ESXi 5.1.
My pool consists of 6 RAW-LUNs with 2GB each on a RAID-Z2. I have a LOG SSD and and a L2ARC SSD both RAW-LUNs.
The VM has 16GB RAM allocated.

zpool import says the following:

Code:

[root@nas ~]# zpool import                                                      
   pool: tank                                                                   
     id: 2599985722204748768                                                    
  state: UNAVAIL                                                                
 status: The pool was last accessed by another system.                          
 action: The pool cannot be imported due to damaged devices or data.            
   see: http://www.sun.com/msg/ZFS-8000-EY                                      
 config:                                                                        
                                                                                
        tank                                            UNAVAIL  missing device 
          raidz2-0                                      ONLINE                  
            gptid/33c13c4e-49ae-11e2-9a7f-000c294ced7d  ONLINE                  
            gptid/3439edca-49ae-11e2-9a7f-000c294ced7d  ONLINE                  
            gptid/34a6920f-49ae-11e2-9a7f-000c294ced7d  ONLINE                  
            gptid/3512b5e0-49ae-11e2-9a7f-000c294ced7d  ONLINE                  
            gptid/3584f095-49ae-11e2-9a7f-000c294ced7d  ONLINE                  
            gptid/3603142b-49ae-11e2-9a7f-000c294ced7d  ONLINE                  
        cache                                                                   
          dsk/da7p2                                                             
        logs                                                                    
          da7                                           ONLINE        

Let me know if you need more info.

I hope you can help me.

Regards,
Nik

I think your issue was not caused by the upgraded, it just manifested after the reboot.
Please open a thread in the Help forum so we can sort this out.

 

[Nik]

Well in general I would agree, but I rolled back to v8.3.0-p1 (old VM) and I can perfectly access the pool. No errors shown.
Then I upgraded again, and ran into the same error.
Now I'm back again to v8.3.0 and everything is working good.

So the question is now, what is wrong with my pool or what is wrong with the 8.3.1 if it is not my pool?

On the first view I would say it has to do with my da7 drive which is partitioned to da7p1 which is my LOG and da7p2 which is my L2ARC.
Does maybe the import fail because of that?

Regards,
Nik

 

[William Grzybowski]

Woah, thats interesting...

It is really hard to tell why this is happening, the freebsd version is basically the same, except for a few patches that are not related to ZFS whatsoever...
But yes, it does look like some issue with your cache/log device, although I can say why it does not happen with 8.3.0... maybe because with the 8.3.1 the hostid changes and it tries to import with -f.

Maybe you could try to remove them before upgrade, and add them back after?

 

[Nik]

Maybe you could try to remove them before upgrade, and add them back after?

Yes, that's my plan, but I want safe my data somewhere first.

I'll post the result tomorrow or on the weekend.

 

[hervon]

Upgraded from 8.3.0. All was done in 5 minutes. Everything is operational. Thanks again!

 

[cyberjock]

In regards to the comments Nik has made...

Every so often someone shows up in the forum and makes a similar claim as Nik. Everything was fine, then they upgrade and the zpool doesn't want to mount. We've been dismissing these out of hand since there really shouldn't be any relationship between upgrading FreeNAS and the zpool going from fully functional and working to no longer mounting. I'm really wondering if we have been improperly dismissing these people out of hand and there really is an obscure bug that we aren't aware of. It may be with ZFS, it may be FreeNAS code, I don't know. It just makes me wonder about all those people that have upgraded and suddenly the zpool says the devices are damaged.

Depending on how this all plays out when Nik makes his new thread, I'm wondering if we should make a recommendation to remove any LOG and CACHE devices attached to zpools before upgrading in the future. Obviously we aren't going to start making this recommendation without observing this behavior more, just my brain wondering what is going on.

 

[Daisuke]

I successfully upgraded from 8.3.0 to 8.3.1, without any zpool issues mentioned above.
With my hardware, I noticed:

• Freenas does not support crash info, /etc/rc warning is OK
• Not sure if the nss-mdns "noise" should be ignored, see ticket 965 > 928
• The AHCI timeouts are still present, reverting to 8.0.4 eliminates them
• The SSD cache status reports as NULL in GUI but OK in console, see ticket 2095

Looking forward to any comments from the devs! :)

Code:

# zpool status
  pool: nas
 state: ONLINE
  scan: scrub in progress since Sat Mar 23 01:13:09 2013
        330G scanned out of 3.65T at 208M/s, 4h40m to go
        0 repaired, 8.82% done
config:

        NAME                                            STATE     READ WRITE CKSUM
        nas                                             ONLINE       0     0     0
          raidz2-0                                      ONLINE       0     0     0
            gptid/3cf03778-1dc5-11e1-8396-002590382e1e  ONLINE       0     0     0
            gptid/3d98e472-1dc5-11e1-8396-002590382e1e  ONLINE       0     0     0
            gptid/3e44f17a-1dc5-11e1-8396-002590382e1e  ONLINE       0     0     0
            gptid/3ef3f815-1dc5-11e1-8396-002590382e1e  ONLINE       0     0     0
            gptid/3f9211d5-1dc5-11e1-8396-002590382e1e  ONLINE       0     0     0
            gptid/4033d56c-1dc5-11e1-8396-002590382e1e  ONLINE       0     0     0
        cache
          gptid/41575359-1dc5-11e1-8396-002590382e1e    ONLINE       3  114K     0

errors: No known data errors

# tail -n 20 /var/log/messages
Mar 22 21:08:14 pluto kernel: GEOM_ELI:     Crypto: software
Mar 22 21:08:14 pluto root: /etc/rc: WARNING: Dump device does not exist.  Savecore not run.
Mar 22 21:08:15 pluto root: /etc/rc: WARNING: failed precmd routine for vmware_guestd
Mar 22 21:08:16 pluto ntpd[2692]: ntpd 4.2.4p5-a (1)
Mar 22 21:08:25 pluto ntpd[2693]: time reset +0.841781 s
Mar 22 21:08:29 pluto avahi-daemon[3113]: WARNING: No NSS support for mDNS detected, consider installing nss-mdns!
Mar 22 21:24:02 pluto ntpd[2693]: kernel time sync status change 2001
Mar 23 01:02:21 pluto kernel: ahcich0: Timeout on slot 24 port 0
Mar 23 01:02:21 pluto kernel: ahcich0: is 00000000 cs 01000000 ss 01000000 rs 01000000 tfd 40 serr 00000000 cmd 10009817
Mar 23 01:03:12 pluto kernel: ahcich0: Timeout on slot 24 port 0
Mar 23 01:03:12 pluto kernel: ahcich0: is 00000000 cs 01000000 ss 00000000 rs 01000000 tfd 50 serr 00000000 cmd 10009817
Mar 23 01:04:02 pluto kernel: ahcich0: Timeout on slot 24 port 0
Mar 23 01:04:02 pluto kernel: ahcich0: is 00000000 cs 01000000 ss 00000000 rs 01000000 tfd 50 serr 00000000 cmd 10009817
Mar 23 01:04:02 pluto kernel: (ada0:ahcich0:0:0:0): lost device
Mar 23 01:04:02 pluto kernel: 
Mar 23 01:04:52 pluto kernel: ahcich0: Timeout on slot 24 port 0
Mar 23 01:04:52 pluto kernel: ahcich0: is 00000000 cs 01000000 ss 01000000 rs 01000000 tfd 50 serr 00000000 cmd 10009817
Mar 23 01:05:30 pluto kernel: ahcich0: Poll timeout on slot 24 port 0
Mar 23 01:05:30 pluto kernel: ahcich0: is 00000001 cs 01000000 ss 00000000 rs 01000000 tfd 450 serr 00000000 cmd 10009817
Mar 23 01:07:25 pluto su: floren to root on /dev/pts/0

 

[joeschmuck]

Installed 8.3.1-Release as a new installation from the ISO image (CD-R). Only found one issue, the drive serial numbers are not displayed under View Disks. Otherwise it appears to operate fine.

 

[HolyK]

Updated from 8.3.0 RELEASE x64 to 8.3.1 RELEASE x64 via GUI update. No problems with update, both ZFS and UFS are fine.

I had some problems with updating plugin jail (successfully updated after all), but i was not able to update Transmission. No error message in console, only one message in GUI in green box like "Some problem occurred". I did not want to spend whole day with haunting the cause, so i removed the old Transmission and i installed new version again. Configuration is like 2min thing so not a big deal.

Thanks!

EDIT: Hmmm, write speed on ZFS pool over CIFS dropped from approx 95-105 Mb/s to 60Mb/s ... i have no idea why ... will need to check it more what happened :/

 

[Daisuke]

Hmmm, write speed on ZFS pool over CIFS dropped from approx 95-105 Mb/s to 60Mb/s ... i have no idea why ... will need to check it more what happened :/

You are not the only one, I had the same issue while upgrading way back from 8.0.2 to 8.0.4. I learned to live with it. :)

 

[William Grzybowski]

There were changes related to CIFS along the path. The most important is AIO. It helps in some cases and degrade performance in others. You can try toggle the AIO in CIFS settings, and tweak the AIO rsize/wsize.

 

[HolyK]

Well ... i forgot that i had scheduled test-backup which started few minutes after FreeNAS upgrade, so that was the cause of slower transfer speed. Anyway i tested again and the speed was still lower than usual (approx 80Mb/s). After some reading and digging, i removed "SO_RCVBUF" and "SO_SNDBUF" from my CIFS Auxiliary parameters and after restarting the CIFS i got again approx 95-100 Mb/s (Different, not-cached transfer offcourse).

Unless you're on a very old Linux or other Unix the kernel is far better off figuring out that itself.

So my aux params are now:

Code:

socket options = TCP_NODELAY IPTOS_LOWDELAY SO_KEEPALIVE
getwd cache = yes

And i have AIO enabled with value "4096" for both min/max.

BTW: There is no need to specifying "read raw = yes, write raw = yes, dead time = 15 , max xmit = 65535" as someone is advising here on forum since these values are already configured by default. Unless you have for some reason different values in "/etc/local/smb.conf", do not specify, its useless :]

Also if someone is using 10G connection in relation with Win7 or Win Server 2008, you may be interested in enabling "SMB Large MTU". It is disabled by default in some versions of Win, so check these registry values and change/reboot in case of need.

DisableLargeMtu=0 (DWORD) // Large MTUs ENABLED
DisableLargeMtu=1 (DWORD) // Large MTUs DISABLED

Windows Server 2008 R2
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters
DisableLargeMtu=0

Windows 7
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\LanmanWorkstation\Parameters
DisableLargeMtu=0

Source: http://technet.microsoft.com/en-us/library/ff625695(v=ws.10).aspx
// Note that the "Setting name" is not the "Key Name", so its a bit confusing on that microsoft library page.

 

[cyberjock]

Yet another example that the best of us can see performance issues with improper tuning. :P
There's a reason why I got rid of all the tuning I had tested extensively in mid-2012 and only have the tuning parameters from Autotune. It seems like its much harder for those of us who aren't FreeBSD pros getting tuning that does work than one would think.

So many people have posted that they've had to remove tuning to get performance back that its almost laughable, myself included.

 

[HolyK]

Thing is that "SO_RCVBUF" and "SO_SNDBUF" gived me a slightly better transfer speed on 8.3.0 or maybe it was something else and this was only "dummy" thing. I never had the rest of the params specified, but when i was looking for some information i found lot of posts recommending these values. I've checked my smb.conf and i found that they are already here, so duno why i should set this manually. Probably because they were different in previous versions of FreeNAS/FreeBSD.

Anyway i need to agree that no crazy tuning is needed since FreeNAS is well-good configured by devs. And IF someone really want to change something, he/she should ALWAYS read WHAT the parameter/value is doing and WHY he want to change it. NOT because "someone" posted it on forum... :)

 

[cyberjock]

Anyway i need to agree that no crazy tuning is needed since FreeNAS is well-good configured by devs. And IF someone really want to change something, he/she should ALWAYS read WHAT the parameter/value is doing and WHY he want to change it. NOT because "someone" posted it on forum... :)

Those are words of wisdom. The problem always seems to be "How far down the rabbit hole can you go?" If you don't know how far "down" you have to go you might think you understand what you are doing and that those parameters/values make sense, bu they don't. Personally, I refuse to just add tweaks without fully understanding how and why it works. Even then it gets very difficult to determine "if" the tweak really helped or if it was just a fluke. There's so many variables with network speeds its hard to rule them all out and prove that your tweak really made the difference. That's why I've given up on tweaking. The defaults provided with FreeNAS seem to work the best and there's more ways to get tweaks wrong than get them right. :)

 

[joeschmuck]

I just turned on Autotune Tunables and my read speed jumped from ~30MB/sec to ~65MB/sec and write speed jumped a little from ~72MB/sec to 80MB/sec. I've never used the tunables before so I'm happy with the results. Now I'll look into the parameters HolyKiller mentioned to see what the potential improvement is.

EDIT:
HolyKiller, Thanks for posting that info. The combination of using the autotune and the CIFS aux settings has significantly boosted performance up to reading ~90MB/sec and writing 107MB/sec. Since my last speed check I have made a few changes to my system such as adding a 6th hard drive and rebuilding the pool to a RAIDZ2, increased my RAM to 16GB and slowed it's speed down from DDR3 1600 to DDR3 1066 (due to stability problems), and of course updating to 8.3.1-Release. You contribution was invaluable to me. Thank You!

 

[Simon00]

Since Autotune is disabled by default on FreeNAS 8.3.1, so I just keep it that way.

Just wondering if Autotune will help on systems w/ only a single core 64-bit cpu & merely 2GB of ram???

They are cheap/low power eMachines, not worth upgrading, really... Use w/ zfs, raid1, 2 hdd, and windows share, 2 user at most, that's all. Was "afraid" to test on the system I had on hand....

If Autotune is so great, I would think and many would appreciate it if it would be ENABLED by default on the next minor release.