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

levak said:
I would like to move my web data from iscsi to NFS, but I'm having some NFS performance issues.

Speeds are ok in both, iSCSI and NFS, but file creation/deletion on NFS are sooooooo BAD:)

On iSCSI I get:
File creation: 11000/s
File deletion: 12000/s
File read: 50000/s

NFS with sync set to disable:
File creation: 2300/s
File deletion: 1600/s
File read: 5000/s

I would expect NFS to be equal to iSCSI, since I disabled the sync option for NFS. Is it enough to just disable sync in nappit or do I have to do something else?

lp, Matej
Click to expand...

Disabling sync is not recommended with NFS, use other tuning options beside tests.

read:
http://constantin.glez.de/blog/2010...ove-oracle-solaris-zfs-filesystem-performance
http://www.solarisinternals.com/wiki/index.php/ZFS_Evil_Tuning_Guide
http://www.solarisinternals.com/wiki/index.php/ZFS_Best_Practices_Guide

Gea
 
Gea, seeing that you are so apt at responding to questions (thank you!), rather than PM you, hopefully you can help me a solve a VMWare issue.

My issue is happening to at least one more person than I and let me quote them as they have already explained the issue quite well:

PigLover said:
When running under ESXi, starting the SE11 VM became unreliable when it was assigned more than 2 virtual CPU cores. I do not totally understand this. It might have been related to the passthrough of the M1015 (I never tried it without it because it was not an interesting configuration). When assigned 3 cores SE11 would begin starting, crash, try again, maybe crash again, and then eventually start up. With 4 virtual cores it would just hang forever on about 1 in 3 restarts. Once it got started it was rock-solid stable even under stress-tests - whatever was happening simply affected starup.
Click to expand...

For reference, my setup is as follows:
Xeon E3 1230
Supermicro X9SCM-F
16GB ECC DDR3
2x IBM M1015 (LSI 9240-8i)
ESXi 4.1 U1
OpenIndiana b148

The same issue that is happening to PigLover is happening to me. Ideally, I would like to use 4 vCPUs on the VM for best performance. Note that with the Xeon 1230 and hyperthreading, I have 8 logical cores. Any help that you can provide to solve this issue is much appreciated so thank you in advance.
 
unclerunkle said:
Gea, seeing that you are so apt at responding to questions (thank you!), rather than PM you, hopefully you can help me a solve a VMWare issue.

My issue is happening to at least one more person than I and let me quote them as they have already explained the issue quite well:



For reference, my setup is as follows:
Xeon E3 1230
Supermicro X9SCM-F
16GB ECC DDR3
2x IBM M1015 (LSI 9240-8i)
ESXi 4.1 U1
OpenIndiana b148

The same issue that is happening to PigLover is happening to me. Ideally, I would like to use 4 vCPUs on the VM for best performance. Note that with the Xeon 1230 and hyperthreading, I have 8 logical cores. Any help that you can provide to solve this issue is much appreciated so thank you in advance.
Click to expand...

Define unreliable. What are your logs saying?

Are you running encryption to take advantage of the built-in AES-NI acceleration? If not (and I don't even know if AES-NI acceleration will even benefit from multiple cores), there's really no need for more than a single core. Save your cores for other VMs. I'm running the same HBAs as you (and it has been stable for months), but I haven't utilized more than a single core with my system. Even with ZFS folder level encryption active on my most heavily used folder, I've maxed the theoretical speed limits of my hard drives. I'm running a Core i7 860. Doing a top on my ZFS server while transferring a large file or lots of small files usually is around 90% idle at the worst.
 
At one point, I had OpenIndiana fully installed, but I have been able to narrow it down to even happening to the boot CD when it first loads. I tried the server (text) version of the OpenIndiana install CD and the same thing happens. It just stalls here:



So, this is in no way related to how OpenIndiana is configured. With only 2 vCPUs in VMWare (all of the other settings unchanged) the install CD boots up just fine. One thing to also note is that CPU usage for the entire host maxes out after it hangs. VM is pulling 9593 mhz according to ESXi.

However, Emulsifide, you do bring up a good point. Do I really need more than 2vCPUs? Not sure yet. I could do more testing as far as that goes, but I would like to be able to have the option if I need it. Also, I'm sure others would like to have the option if need be. Plus, my curious side like solving issues :p
 
Last edited:
I think I found a bug in napp-it. You can't create hidden smb shares. If you want to name a share something like sharename$, $ characters gets removed... It works OK from command line.

lp, Matej
 
about ESXi and 4 vCPU for OI

I have had also sometimes (not always) boot problems with 4 cores assigned to OI.
There was also an advice from the german vmware-forum not to assign 4 cores to ESXi guests at all.

So i also suggest to assign 2 vCPU to each guests.
RAM is more critical, CPU is mostly a problem with encryption and a
smaller problem with compress, raid-z and higher levels of checksums
(For ESXi storage, Raid-10 is suggested, fastest option with lowest CPU demands)

@levak
I will fix that in next nightly

Gea
 
Can someone help me. I just installed Nexenta CE 3.0.5 and i am not sure if I am creating my pool correctly. I don't see an option to create a RaidZ pool. See screen shot below

help.png
 
Vengance_01 said:
Can someone help me. I just installed Nexenta CE 3.0.5 and i am not sure if I am creating my pool correctly. I don't see an option to create a RaidZ pool. See screen shot below
Click to expand...

i suppose, you have build a pool from 3 basic disks (1TB?) with a hotfix
- without any redundancy- (like a Raid-0)

-> When you create a new Volume (When Nexenta is talking about a Volume, they mean a ZFS pool)
or add disks (= add a vdev), you can select the redundancy type of your added disks (vdev)


Gea
 
Last edited:
Okay I think I have figured it out. You need to selected all the drives, and then select your redundancy type
 
So I got an ISCSI connection up and created a 100GB Slice. Network speeds are not bad at about 65MB/s, but I am getting the yo-yo effect where it hits 80-90% then down to 5-10% and then back up. On the machine I am copying data from, it has a dedicated Intel GT 1000 PCI-E card. The HD I am using is a WD 640GB Blue so I would think its capable of 80-100MB/s. When making a single copy, my CPU usage is in the 80-90% range. Granted I know the HP Micro server's 1.3GHZ Neo 2 chip is not the fastest, but I with 4GB ram, I figured that would be enough to saturate a 1GB connection. This is with a completely stock fresh install of Nexenta CE 3.0.5. I would love any tips or tuning suggestions.
 
unclerunkle said:
For reference, my setup is as follows:
Xeon E3 1230
Supermicro X9SCM-F
16GB ECC DDR3
2x IBM M1015 (LSI 9240-8i)
ESXi 4.1 U1
OpenIndiana b148

The same issue that is happening to PigLover is happening to me. Ideally, I would like to use 4 vCPUs on the VM for best performance. Note that with the Xeon 1230 and hyperthreading, I have 8 logical cores. Any help that you can provide to solve this issue is much appreciated so thank you in advance.
Click to expand...
Does this setup work with a straight install of OpenIndiana? (no ESXi involved) I'm looking to get a near identical setup, but am worried about support for the new C204 chipset. I don't plan to use ESXI
 
_Gea said:
about ESXi and 4 vCPU for OI

I have had also sometimes (not always) boot problems with 4 cores assigned to OI.
There was also an advice from the german vmware-forum not to assign 4 cores to ESXi guests at all.

So i also suggest to assign 2 vCPU to each guests.
RAM is more critical, CPU is mostly a problem with encryption and a
smaller problem with compress, raid-z and higher levels of checksums
(For ESXi storage, Raid-10 is suggested, fastest option with lowest CPU demands)

@levak
I will fix that in next nightly

Gea
Click to expand...

Thank you for the clarification. I will stick to only 2 vCPUs then.
 
zeroARMY said:
Does this setup work with a straight install of OpenIndiana? (no ESXi involved) I'm looking to get a near identical setup, but am worried about support for the new C204 chipset. I don't plan to use ESXI
Click to expand...

Check out this thread.

I am not home currently, but I could test. Looks like their may be compatibility issues directly though.
 
unclerunkle said:
Check out this thread.

I am not home currently, but I could test. Looks like their may be compatibility issues directly though.
Click to expand...

If it's not too much trouble some confirmation would be nice :) After some additional reading ESXi does now sound like a good idea thought...
 
Last edited:
General ZFS question: My plan was initially to fill up a 24-bay in Norco in 3 phases. Start with 8 2TB Hitachi hard drives and a 2x SAS port LSI card in a single vdev with RAIDZ2. Then 6 months later add another SAS card and 8 more hard drives. Then 6 months after that add the third SAS card and the final group of 8 hdds. However, it seems vdev's can't be expanded, so each group have to have it's own vdev and RAIDZ2 parity. So that means I would lose 6 drives worth to parity. That's just too much for me. So instead I should do 2 vdevs of 12 drives, so that I would only lose 4 drives worth to parity with RAIDZ2. Or maybe I should just do RAIDZ1 with my original plan of 3x 8 drive groups. I will only be storing Blu-ray rips, so if a drive fails, I'll keep the server off until I get a replacement.

Is my information correct? Thoughts?
 
zeroARMY said:
General ZFS question: My plan was initially to fill up a 24-bay in Norco in 3 phases. Start with 8 2TB Hitachi hard drives and a 2x SAS port LSI card in a single vdev with RAIDZ2. Then 6 months later add another SAS card and 8 more hard drives. Then 6 months after that add the third SAS card and the final group of 8 hdds. However, it seems vdev's can't be expanded, so each group have to have it's own vdev and RAIDZ2 parity. So that means I would lose 6 drives worth to parity. That's just too much for me. So instead I should do 2 vdevs of 12 drives, so that I would only lose 4 drives worth to parity with RAIDZ2. Or maybe I should just do RAIDZ1 with my original plan of 3x 8 drive groups. I will only be storing Blu-ray rips, so if a drive fails, I'll keep the server off until I get a replacement.

Is my information correct? Thoughts?
Click to expand...

Honestly, it doesn't sound like the info is all that important, but how are you planning to back all of that up?

I would just use raidz1 and keep the space. If it was mission critical then I would rethink the array into striped mirrored vdevs or even just a 3 way mirror, but for blu ray rips that I am assuming you still have the discs for and could re-rip them, that kind of redundancy isn't necessary.
 
Hoping for some easy help. So I got napp-it installed and must have blown it on the password setup because now I can't login with the default password which the napp-it homepage says should be blank. I searched for the napp-it.cfg file but couldn't find it anywhere. I'm running "SunOS fileserver 5.11 NexentaOS_134f i86pc i386 i86pc Solaris".

Can anyone help a napp-it newbie get access back.

Thanks in advance
 
wildman72 said:
Hoping for some easy help. So I got napp-it installed and must have blown it on the password setup because now I can't login with the default password which the napp-it homepage says should be blank. I searched for the napp-it.cfg file but couldn't find it anywhere. I'm running "SunOS fileserver 5.11 NexentaOS_134f i86pc i386 i86pc Solaris".

Can anyone help a napp-it newbie get access back.

Thanks in advance
Click to expand...

delete /var/web-gui/data/napp-it/napp-it.cfg
to get defaults


Gea
 
no reboot is not needed.
just login without pw.

(deleted settings are only those from menu napp-it-setup)

Gea
 
Today, when I was erasing some unused datasets I realized there would be nice to have some 'oops I accidentally erased the wrong dataset' protection. While thinking of any good idea, a came to the idea where you would click on the dataset you wanted to erase and then a popup window would show and you should type the name of the dataset you are erasing, again. Becase if you are in a hurry, you can click OK without looking at what dataset you are erasing....

What do you think?

Matej
 
Mine isn't working. As a work-around I've installed msmtp with cron scubs and such--seems to work fine.
 
_Gea,

I'm testing out OI and napp-it in a Virtualbox install before I migrate my 20TB WHS to ZFS. I'm loving it more than ZFSGuru right now. There is so much more to it, and it helps me learn quite a bit both about ZFS and Solaris.

I don't run a domain at home, but I have 2 PC's with identical login credentials. This is for administrator access to WHS shares. I'm trying to set up something similar with napp-it, but I can't quite figure it out. I create a smb user with the same username/password as I am using on my 2 client machines. I add the smb user to the Administrators group. For the folders I want to share, I have guest enabled. I set 755 for permissions as well as "Everyone" = read_set. When I type in my \\servername in Windows, I get a login prompt. If I input my Bonz account, I have readonly access. If I input anything, I have readonly access. Only if I put in the root account do I have full access.

Here are 2 screenshots. Maybe someone can explain what I'm doing wrong. What I'm trying to accomplish is read only access for everyone, and full access for myself over the network.

ZFS_Test_Share_04252011.png

ZFS_Test_User_04252011.png
 
Last edited:
BonzTM said:
What I'm trying to accomplish is read only access for everyone, and full access for myself over the network.
Click to expand...

You can set ACL either by folder/file permissions or via share permissions.

1. via folder permissions
re-set basics:
- - disable guest access
- - set permissions to 777
- - set share permissions to full for everybody

-re-set root pw once after installation of napp-it (to create a smb pw too)
-smb-connect from windows as user root
-set acl from Windows as desired
- - everybody=read
- - root = full
- - user1,user2 or a group=full or modify (full=modify + allowed to change permissions)


2. via share permissions (need a smb service restart after changing)
-smb-connect from Windows as a user who is a member of Solaris smb administrators
-start Windows-computer-management
-connect computer-management to Solaris
-set share permissions via computer-management as desired
- - root=full
- - user1, user 2 or a group=full or modify
- - everybody=read

-> folder permissions must be a least the same !!


3. a combination of both


4. about napp-it ACL-settings in menu folder
--> only basic settings; set special settings always from Windows (easier)


5. If you want to set ACL as another user xx (not root, root is owner)
you have to set a mapping Windows user xx = unixuser root


6. about unix perm vs ACL
It is not a either-or. ACL permissions are based on unix permissions.
if you reduce ACL, unix permissions are reduced automatically to be at the same level and vice versa !!
If you use ACL, start always with unix permissions 777 and set the needed ACL


Gea
 
Last edited:
I think command
idmap add winuser:Administrator unixuser:root
would do the trick(I hope the syntax is correct).

Matej
 
Thanks Gea, I will reset everything and give it a shot. Also, do you know if there is a way in OI to auto-login? This will run mostly headless unless I need to attach peripherals for some reason. I would like the Gnome GUI to do a few things that I'm not comfortable with in the CLI, so I opted for the Desktop version.
 
levak said:
I think command
idmap add winuser:Administrator unixuser:root
would do the trick(I hope the syntax is correct).

Matej
Click to expand...

in the above example, you need a:
idmap add winuser:Bonz unixuser:root

or you have to create a smb-user administrator and
connect as smb-user administrator (Windows alike)

but the easiest way is always:
forget mapping, use root to set top-level ACL (initial set only, you can then set Bonz=full)
(unless the are security concerns about other services with user root)


Gea
 
BonzTM said:
Thanks Gea, I will reset everything and give it a shot. Also, do you know if there is a way in OI to auto-login? This will run mostly headless unless I need to attach peripherals for some reason. I would like the Gnome GUI to do a few things that I'm not comfortable with in the CLI, so I opted for the Desktop version.
Click to expand...

i would use desktop sharing via a vnc client like tightvnc (Windows)
(enable desktop sharing in OI/SE11 GUI/live version with menu system-setting-desktop sharing

Gea
 
_Gea said:
i would use desktop sharing via a vnc client like tightvnc (Windows)
(enable desktop sharing in OI/SE11 GUI/live version with menu system-setting-desktop sharing

Gea
Click to expand...

Yea that's what I'm doing, however, in order for that to run after a reboot I have to go to the machine, and log in manually. I'd like it to auto-login for me, then I can manage it via VNC headless.
 
Last edited:
BonzTM said:
Yea that's what I'm doing, however, in order for that to run after a reboot I have to go to the machine, and log in manually. I'd like it to auto-login for me, then I can manage it via VNC headless.
Click to expand...

I SSH in and start vncserver then I can VNC in.. I guess that way it's also more secure so VNC isn't always open... but I know what you mean. I'm too newb to set it up to allow vnc before I even login to the computer itself...
 
Hi all, can anyone explain me the differences between (and possible cause):

Illegal Request
Soft Errors
Hard Errors
Transport Errors

I've found (in napp-it, under disk-controller) some disks with info like this:

Illegal Request: 97 Predictive Failure Analysis: 0
c9t1d0 Soft Errors: 166 Hard Errors: 0 Transport Errors: 0
Vendor: ATA Product: SAMSUNG HD204UI Revision: 0001 Serial No:
Size: 2000.40GB <2000398934016 bytes>
Media Error: 0 Device Not Ready: 0 No Device: 0 Recoverable: 166

thank you
 
gea:

is it possible to monitor the temperature of the hard drives from napp-it?

also monitoring the cpu temp would be nice

thanks!
 
Not sure if this will help or not, but here's what I have in Nexenta for gmail notifications:
SMTP server: smtp.gmail.com
SMTP user: your_email_address (eg, [email protected])
SMTP password: your_password
SMTP Send Timeout: 30 seconds
SMTP Authentication: SSL
email addresses: email you want notifications sent to
from email address: address you want to show up in the from field of the notifications
 
_Gea, I have been following your thread for a while and would like to thank you for the work you have put into this product. You're providing a service at a level many commercial companies should strive for. Cheers to you.
 
You must log in or register to reply here.
Back
Top