Issues with metaSAN

Hi-

We are beta testing the new Sonnet Technologies VFibre shared storage system:

http://www.sonnettech.com/news/pr2010/pr091010_fusionrx1600vfibre.html

It is a fiber and ethernet based SAN with a windows server built in to a 48TB fibre channel raid (using SAS internally) using metaSAN as the metadata server and windows computer as the master. We've had it for about 2.5 weeks now. I hope I explained all that correctly, I am new to the SAN admin world.

Right now we have two metaSAN licenses (desktops connected via fibre), and run between 1-3 metaLAN licenses (ethernet). These are all running on Macs, the windows server is NTFS, of course.

Everything was going pretty well until this past weekend when without warning, the San disappeared from everyone's desktop, knocking all files offline in FCP. The volume is still listed as mounted on the windows server, but the SAN is "Not Available" according to metaSAN. I then have to go in to metaSAN and add the volumes back to both metaSAN and then redefine the metaLAN volume as well. Everything then mounts back up, and away we go.

No data loss or corruption of any sort, but it is a bit disconcerting that the volumes keep getting "lost" on the master. This has now happened two more times since Sunday for a total of three times in a week. Here's a log from a laptop based ethernet connection:

08/17/11 16:26 Volume "HighInFibre" was removed from volumes list.
08/17/11 16:26 Failed to process master changed notification for volume "HighInFibre".
08/17/11 16:26 Volume "HighInFibre" was unmounted from client.
08/17/11 16:26 Volume "HighInFibre" was removed from volumes list.
08/17/11 16:26 Master changed notification received for volume "HighInFibre".
08/17/11 09:04 Volume "HighInFibre" was mounted as client.
08/17/11 09:04 Volume "HighInFibre" was added to volumes list.
08/17/11 09:04 Volume "HighInFibre" was added to volumes list.
08/17/11 09:04 Currently logged system user set successfully.
08/17/11 09:04 Trying to set currently logged system user...

I also have logs from the master if you need them. Teamviewer is also installed on the server in case you want to take a look.

Any ideas on what is going on here? I have also been working with Sonnet tech support, but my contact had to go away on emergency so I figured I would try here until he is able to respond.

Thanks so much,

Jeremy

[Brandon Kraemer] "I would certainly suggest you contact Tiger with support logs generated from the metaSAN interface. "

Will do.

It just happened again. You know what is weird, it seems to happen at around 4:30PM all the time. I have no idea if that's just coincidence, or what.

[Brandon Kraemer] "1. we reformatted our sonnet volumes."

Yikes. That would be no good. They are brand new.

[Brandon Kraemer] "2. in SAN Managment... Advanced... Master Priorities, we set the master IP to priority High and all other client nodes to priority None."

Yes, that is already setup that way. Thanks.

What about the "Data Masters" button? It appears all the metaSAN clients are setup for masters. That's normal, right?

Jeremy

Thanks so much, Brandon.

Jeremy

[Bernard Lamborelle] " You mentioned this happening twice in a row around 4h30 and only recently. Do you have any special processes or activities taking place at that moment (i.e. anti-virus, defrag or backup application that could interfere)?"

Not that I know of. The windows part of the machine and the drives come preformatted and ready to go. The only thing to do is get the DNS setup, and define everything in metasan. I will have to do some more digging on that. Unfortunately, windows is very foreign to me.

[Bernard Lamborelle] "But if you have not been making lots of changes to your SAN Definition, it could also be something else."

No. It's stayed the same. What is weird is that the drive stays mounted to the VFibre during this, it's just the SAN that gets booted.

It has now happened 4 times. From memory, three of those times were right around 4:30PM (16:30).

It was Sunday, Monday, Wednesday, Thursday.

Another thing I should mention, the first time it happened, one of the drives in the raid6 config reported as faulted, it's odd as we only had the machine for barely two weeks at that point. I rebuilt the raid, and hasn't faulted since then.

Thanks for your response.

Jeremy

Jeremy

Awesome.

Thanks so much!

Jeremy

Hi Jeremy,

I was hoping you could give us an update on how things have been going since the config of your system was redone. Is it too early to tell yet?

Cheers,

Bernard

Bernard Lamborelle
bernard at tiger-technology dot com
http://www.tiger-technology.com
514-667-2015

Hi Jeremy,

It's been a while since we heard from you...

Has your system been rock solid? Did you manage to fix the "few other weird things"?

Cheers,

Bernard

Bernard Lamborelle
bernard at tiger-technology dot com
http://www.tiger-technology.com
514-667-2015

HII ,,,

Do you need to metaSAN two Ethernet networks such as the Xsan , To dedication one of the private (metadata) network and one for the public ??

but Xsan Create a private network with open directory users ( ACL Permissions ) Is there a special configuration of the metaSAN?