Troubleshooting

Troubleshooting :: RE: Xsan config file lost, any way to recover SAN volume?

lotte — Wed, 08 Sep 2010 21:21:55 GMT

Author: lotte
Posted: Wed Sep 08, 2010 4:21 pm (GMT -5)

Hi thequietman44, here comes a .cfg file which I think of could work, but there are a few things which came on my mind while creating it, so maybe one of the other guys reading this can bring a little bit light in my darkness!

What will happen if the names of the stripegroups are different from the original ones?
What will happen if the Affinities are spelled wrong?
I know that if FsBlocksize and/or StripeBreadth differ from the original config the Filesystem will not come up, but will not be affected.
I�ve set the write parameter for all StripeGroups to disable so that if hopefully the Volume comes up you will only be able to read from it and are not allowed to write for safety reasons. In any case we have luck with your Volume I would recommend to backup all data then create a new Volume.

What else I don�t know is the name of your Volume which is very important, so first you need to know the exact spelling of the Volume.

Here comes the config:
------------------------snip--------------------------

FileLocks Yes
GlobalSuperUser Yes
Quotas Yes
WindowsSecurity No
ForceStripeAlignment Yes
Debug 0
AllocationStrategy Round
InodeExpandMin 128
InodeExpandInc 128
InodeExpandMax 2048
BufferCacheSize 32M
JournalSize 64M
FsBlockSize 4K
InodeCacheSize 8K
MaxConnections 75
MaxLogSize 10M
ThreadPoolSize 128

[DiskType Meta_Lun]
Sectors 781367296
SectorSize 512

[DiskType Project_Lun]
Sectors 3906926592
SectorSize 512

[DiskType Offline_Lun]
Sectors 2344089600
SectorSize 512

[DiskType FCP_Lun]
Sectors 8790767583
SectorSize 512

[Disk MetadataLUN]
Type Meta_Lun
Status UP

[Disk ProjectLUN]
Type Project_Lun
Status UP

[Disk OfflineLUN]
Type Offline_Lun
Status UP

[Disk FCP_LUN1]
Type FCP_Lun
Status UP

[Disk FCP_LUN2]
Type FCP_Lun
Status UP

[StripeGroup Metadataandjournal]
Status UP
Exclusive Yes
Metadata Yes
Journal Yes
Read Enabled
Write Disabled
StripeBreadth 256
Node MetadataLUN 0

[StripeGroup Project]
Status UP
Exclusive No
Metadata No
Journal No
Affinity Project
Read Enabled
Write Disabled
StripeBreadth 256
MultiPathMethod Rotate
Node ProjectLUN 0

[StripeGroup Offline]
Status UP
Exclusive No
Metadata No
Journal No
Affinity Offline
Read Enabled
Write Disabled
StripeBreadth 256
MultiPathMethod Rotate
Node Offline_Lun 0

[StripeGroup FCP]
Status UP
Exclusive No
Metadata No
Journal No
Affinity FCP
Read Enabled
Write Disabled
StripeBreadth 256
MultiPathMethod Rotate
Node FCP_LUN1 0
Node FCP_LUN2 1

--------------------------------snap------------------------------

And here is what to do:

Switch off mdc2, then in a terminal on mdc1 as root type: (always without the quotes):

"launchctl unload /System/Library/LaunchDaemons/com.apple.xsan.plist" and if that does not work type "Systemstarter stop acfs"
Create a file with the content between "snip" and "snap" and save it to "/Library/Filesystems/Xsan/config/YourVolumeName.cfg" where YourVolumeName is the exact Name of how the Volume was called before!

In the terminal type:
"launchctl load /System/Library/LaunchDaemons/com.apple.xsan.plist" (or if that fails "SystemStarter start acfs")

Then type "cvadmin"
in cvadmin type "start YourVolumeName"
and if that goes well type "activate YourVolumeName"

If that again goes well you maybe lucky by mounting the Volume like:

"mkdir /Volumes/YourVolumeName"
"mount -t acfs YourVolumeName /Volumes/YourVolumeName"

If you see any files and directories you should be able to backup all the SAN content.

If not, what is not unlikely please post the corresponding output of /var/log/system.log and of Library/Filesystems/Xsan/debug/nssdebug.out

I would really appreciate if anybody can have a look over the above and correct me if I forgot about something or made wrong explanations, or else wrote anything stupid!

Anyway, I will have a look in here asap, but I�m on a client side with no internet acces, so it will take some time for me to read...

Good Luck Lotte

Troubleshooting :: RE: Xsan config file lost, any way to recover SAN volume?

morphenine — Wed, 08 Sep 2010 17:31:53 GMT

Author: morphenine
Posted: Wed Sep 08, 2010 12:31 pm (GMT -5)

Quote:

My job now is to see that none of the above mistakes happens again.

Welcome to the club!

I guess I didn't see the part about upgrading to Leopard server, sorry.

I wish I could be of more help to you in recovering the volume, I'm sure if I had a few spare installations and all the time in the world to screw around with it that we could figure it out...and a pet unicorn would be nice too. Unfortunately (fortunately for me), I've got no experience with that aspect of Xsan. I'm sure someone here (like lotte, or others) can help you out more.

However, if you want best practices then I'd be happy to lend a hand. Feel free to ask away if you have any questions. I'd rather [for your sake] see you ask a hundred questions about how to do things right, than one question about how to fix something that you did wrong. Of course, it something goes wrong, thats also why most of the people are here. If we can we'll help.

Good luck recovering, I wish you the best!

Troubleshooting :: RE: Xsan config file lost, any way to recover SAN volume?

thequietman44 — Wed, 08 Sep 2010 13:22:57 GMT

Author: thequietman44
Posted: Wed Sep 08, 2010 8:22 am (GMT -5)

The default values are "FSBlocksize 4K" and "StripeBreadth 256". Output of cvlabel -l is:

Code:

/dev/rdisk4 [APPLE Xserve RAID 1.51] acfs "FCP_LUN1" Sectors: 8790767583. SectorSize: 512. Maximum sectors: 8790767583.
/dev/rdisk5 [APPLE Xserve RAID 1.51] acfs "MetadataLUN" Sectors: 781367296. SectorSize: 512. Maximum sectors: 781367296.
/dev/rdisk7 [APPLE Xserve RAID 1.51] acfs "ProjectLUN" Sectors: 3906926592. SectorSize: 512. Maximum sectors: 3906926592.
/dev/rdisk8 [APPLE Xserve RAID 1.51] acfs "FCP_LUN2" Sectors: 8790767583. SectorSize: 512. Maximum sectors: 8790767583.
/dev/rdisk6 [APPLE Xserve RAID 1.51] acfs "OfflineLUN" Sectors: 2344089600. SectorSize: 512. Maximum sectors: 2344089600.

Morphenine, that time has already come and gone, and no one knows who to slap because no one is responsible for maintaining the SAN. I did find out that several people had asked why the contents of the SAN were never backed up somewhere, but no one had ever thought (or known) to back up the .cfg file.

Believe me, there are already some major changes being made to how the SAN will be used and a backup plan is first on the list. There were a lot of issues with the way it was used, the lack of understanding of how a SAN works, no one to properly maintain it, etc. I'm currently working on a "best practices" proposal that will prevent this and hopefully many other failures from happening in the future, or provide a "Plan B" in case of failure.

Mistake #2 from my original post was the cause of the issue. Both MDCs were "upgraded" from Tiger Server to Leopard and the Xsan software was removed in the process. As I said, no one knew anything about the .cfg files, so there was no concern over blowing away the Xsan software.

If only one controller were upgraded, or if the .cfg file were backed up, or if the data on the SAN were backed up, or if everyone were properly informed on the structure of Xsan, this probably wouldn't have happened. My job now is to see that none of the above mistakes happens again.

Troubleshooting :: RE: Xsan config file lost, any way to recover SAN volume?

morphenine — Wed, 08 Sep 2010 00:42:20 GMT

Author: morphenine
Posted: Tue Sep 07, 2010 7:42 pm (GMT -5)

Its a bit old, but theres also this if it helps:
http://wiki.xsanity.com/groups/xsanitywiki/wiki/8bb1a/Anatomy_of_a_.cfg_file.html

Troubleshooting :: RE: Xsan config file lost, any way to recover SAN volume?

morphenine — Wed, 08 Sep 2010 00:35:28 GMT

Author: morphenine
Posted: Tue Sep 07, 2010 7:35 pm (GMT -5)

I would say that this is probably a good place to stop, take a deep breath, and slap someone upside the head for not making a highly redundant-capable system highly redundant.

Then I would say the number one thing you need to remember is that if there is any hope of recovering this volume, you need to make sure that nothing is written to those RAIDs...especially the metadata lun. I haven't ever done it, but you can rebuild a config file if you really know what you're doing and you have all the info. Unfortunately, I'm not as familiar with Xsan 1 as I am with Xsan 2.
As a semi-last resort, you might be able to take the internal drive of the MDCs to a recovery shop and *possibly* recover the config file (but thats hit-or miss). I've heard theres also services that will recover your Xsan volume for you...for a nominal fee. But you'd have to REALLY REALLY need whats on that volume.

Also, what happened? Did the MDCs take a dive? (Did you check for configurations on both?), Did someone delete the volume? did multiple drives die? How did it get like this?

Bottom Line, whether or not you are able to recover your data, make sure this is an experience you learn from.

Troubleshooting :: RE: Filesystem RAS events undeliverable

tmcnamara — Tue, 07 Sep 2010 23:21:01 GMT

Author: tmcnamara
Posted: Tue Sep 07, 2010 6:21 pm (GMT -5)

Great. So I guess the Filesystem RAS events undeliverable message is the log equivalent of what the kbase article is describing, is that what you're getting at?

Troubleshooting :: RE: Filesystem RAS events undeliverable

abstractrude — Tue, 07 Sep 2010 22:39:38 GMT

Author: abstractrude
Posted: Tue Sep 07, 2010 5:39 pm (GMT -5)

http://support.apple.com/kb/TS3440

Troubleshooting :: RE: Filesystem RAS events undeliverable

tmcnamara — Tue, 07 Sep 2010 21:26:25 GMT

Author: tmcnamara
Posted: Tue Sep 07, 2010 4:26 pm (GMT -5)

The xserves are the 8core nehalem, but yes, the mac pros are the brand new 8 core mac pros. I've seen this once or twice in an xsan with octocore xeons though; this is just very repeatable in this setup.

They are all on 10.6.4 and have 2.2 xsan installed along with the 2.2.1 Snow Leopard Filesystem update.

Troubleshooting :: RE: Xsan config file lost, any way to recover SAN volume?

lotte — Tue, 07 Sep 2010 20:17:11 GMT

Author: lotte
Posted: Tue Sep 07, 2010 3:17 pm (GMT -5)

Congratulation, this seems to be even more complicated than expected, cause there also seem to be folders with affinity...
If you try (just try, not completly do!!!) to create a new Volume what values are given for FSBlocksize and Stripesize, as far as I remember can this be 32 for FSBlocksize and 256 for StripeBreadth?
And can you please post the output of "cvlabel -l" in Terminal. (Has to be run as root!!!)

Lotte

Troubleshooting :: RE: Filesystem RAS events undeliverable

abstractrude — Tue, 07 Sep 2010 18:40:10 GMT

Author: abstractrude
Posted: Tue Sep 07, 2010 1:40 pm (GMT -5)

are they the new mac pros?
also did you do the snow leopard xsan update. and not just the 2.2 disc.

Troubleshooting :: RE: Xsan config file lost, any way to recover SAN volume?

thequietman44 — Mon, 06 Sep 2010 18:46:56 GMT

Author: thequietman44
Posted: Mon Sep 06, 2010 1:46 pm (GMT -5)

LUNs:

MetadataLUN
OfflineLUN
ProjectLUN
FCP_LUN1
FCP_LUN2

Storage Pools:

MetadataPool
OfflinePool
ProjectPool
FCP_Pool

Troubleshooting :: RE: Xsan config file lost, any way to recover SAN volume?

lotte — Mon, 06 Sep 2010 13:48:33 GMT

Author: lotte
Posted: Mon Sep 06, 2010 8:48 am (GMT -5)

It has to be done by hand, I would like to ask all other�s if anyone can remember or even test the default values so that we can provide thequietman44 with a new .cfg file...
What we will need anyway is the name of all Lun�s. can you post them here?

Lotte

Troubleshooting :: RE: Xsan config file lost, any way to recover SAN volume?

thequietman44 — Sun, 05 Sep 2010 19:32:50 GMT

Author: thequietman44
Posted: Sun Sep 05, 2010 2:32 pm (GMT -5)

There is no folder at ~/Library/Xsan or /Library Xsan on the new MDC.

As far as I know, the volume was created with default values, although there is no record of what they actually were. The person who set up the Xsan is long gone and all we have to work with are some sketchy diagrams .

How would I go about recreating the .cfg file? Is that something that needs to be created entirely by hand, or can you have Xsan admin create one with approximately the right values and then modify them to match?

Troubleshooting :: RE: Xsan config file lost, any way to recover SAN volume?

lotte — Sun, 05 Sep 2010 08:32:01 GMT

Author: lotte
Posted: Sun Sep 05, 2010 3:32 am (GMT -5)

Well the main thing is to know the filesystem blocksize and the stripebreadth of each stripegroup. So if anyone can remember what parameters have been used during creating of the Volume there is of course a change to recreate a new .cfg file which can bring up the Volume. Can you have a look at ~/Library/Xsan(admin) what exists there?

Lotte

Troubleshooting :: Xsan config file lost, any way to recover SAN volume?

thequietman44 — Sun, 05 Sep 2010 02:06:36 GMT

Author: thequietman44
Subject: Xsan config file lost, any way to recover SAN volume?
Posted: Sat Sep 04, 2010 9:06 pm (GMT -5)

I realize there may be no hope in this situation considering the number of worst practices in this scenario, but as the only person who knows anything at all about Xsans the task of recovery and rebuilding is mine.

Background:
2 XServe G5s (controllers) with 2 XServe RAIDs running Xsan 1.4.2 with 5 clients. 1 SAN volume made up of 4 pools, each pool has 1 LUN execpt for the last which has 2 LUNs.

Mistake #1: Nothing was backed up. Neither the volume config file nor data stored on the SAN.
Mistake #2: Upgrading both controllers to Leopard Server at the same time.

Result:
RAIDs are online, can log in through RAID Admin and see the arrays.
LUNs show up in Xsan Admin with proper sizes and labels.
Storage pools and volume do not show up, Xsan Admin asks to create a volume.

So, my question is, if the LUNs are intact is there any way to rebuild/recreate the SAN volume with all of its data or is everything hosed? We do have a diagram showing the labels of all the LUNs, pools, affinities, and volume, but no config file.

Thanks for any help you can provide.

Troubleshooting :: RE: Finder crashes after xsan finder copy, Apple is stumped

smlsan — Fri, 03 Sep 2010 16:43:34 GMT

Author: smlsan
Posted: Fri Sep 03, 2010 11:43 am (GMT -5)

Definitely sounds like the issue we have been working on. Let us know if defrag functions as a workaround for now.

FYI - testing in our lab reliably produces hangs with 10,000+ extents. It's less reliable in the 3,000-8,000 range. This is observational only - no specific testing was done to find a particular threshold and it appears to affected by memory use on the client.

I suspect those files with 6,000 extents and above will produce hangs more often than not if you play them in Quicktime or preview/get info.

Troubleshooting :: RE: Finder crashes after xsan finder copy, Apple is stumped

dmastroluca — Fri, 03 Sep 2010 16:30:54 GMT

Author: dmastroluca
Posted: Fri Sep 03, 2010 11:30 am (GMT -5)

I am so excited that there might actually be a workaround for this!! Here is my extent count on just one of my folders. This folder is for long form records of our City Council Meetings, usually 3-4 hours in length @ DV25. These files are the ones that give us the most problems.

sh-3.2# snfsdefrag -c -r /Volumes/KCLV_AIR_OPS/PLAYOUT/Council\ Meeting
/Volumes/KCLV_AIR_OPS/PLAYOUT/Council Meeting/Council PM Pt.mov: 1 extent(s)
/Volumes/KCLV_AIR_OPS/PLAYOUT/Council Meeting/Council AM Pt. 2 080410.mov: 11513 extent(s)
/Volumes/KCLV_AIR_OPS/PLAYOUT/Council Meeting/._Council AM Pt. 2 080410.mov: 1 extent(s)
/Volumes/KCLV_AIR_OPS/PLAYOUT/Council Meeting/Council AM 072110 INGEST.mov: 61 extent(s)
/Volumes/KCLV_AIR_OPS/PLAYOUT/Council Meeting/._Council (CLOSE).mov: 1 extent(s)
/Volumes/KCLV_AIR_OPS/PLAYOUT/Council Meeting/Council AM Pt. 1 090110.mov: 1363 extent(s)
/Volumes/KCLV_AIR_OPS/PLAYOUT/Council Meeting/Council AM 072110 Item 55.mov: 6075 extent(s)
/Volumes/KCLV_AIR_OPS/PLAYOUT/Council Meeting/City Council AM Pt. 2 081810.mov: 12880 extent(s)
/Volumes/KCLV_AIR_OPS/PLAYOUT/Council Meeting/._Council PM Pt. 3 080410.mov: 1 extent(s)
/Volumes/KCLV_AIR_OPS/PLAYOUT/Council Meeting/Council PM Pt. 3 080410.mov: 8248 extent(s)
/Volumes/KCLV_AIR_OPS/PLAYOUT/Council Meeting/._Council PM Pt. 3 081810.mov: 1 extent(s)
/Volumes/KCLV_AIR_OPS/PLAYOUT/Council Meeting/._Special City Council 051810.mov: 1 extent(s)
/Volumes/KCLV_AIR_OPS/PLAYOUT/Council Meeting/Council PM Pt. 3 081810.mov: 9315 extent(s)
/Volumes/KCLV_AIR_OPS/PLAYOUT/Council Meeting/._Council Closed Captioning Announcement.mov: 1 extent(s)
/Volumes/KCLV_AIR_OPS/PLAYOUT/Council Meeting/Council AM Pt. 2 090110.mov: 11156 extent(s)
/Volumes/KCLV_AIR_OPS/PLAYOUT/Council Meeting/._Council AM Pt. 1 081810.mov: 1 extent(s)
/Volumes/KCLV_AIR_OPS/PLAYOUT/Council Meeting/Council AM Pt. 1 081810.mov: 1037 extent(s)
/Volumes/KCLV_AIR_OPS/PLAYOUT/Council Meeting/._Council AM Pt. 1 080410.mov: 1 extent(s)
/Volumes/KCLV_AIR_OPS/PLAYOUT/Council Meeting/Council AM Pt. 1 080410.mov: 1448 extent(s)

I usually use "Get Info" as a way of telling that a hang is happening. The "more info" pane will always say "Fetching" with a spinning gear during the hang. That is when I know I have to do a power cycle to shake it free of the funk.

I am defragging my high extent count files right now!!!!!! Thanks for this info. Could you please contact me via email for a sidebar question? My email is dmastroluca@lasvegasnevada.gov

Dan Mastroluca
Chief Engineer
KCLV-City of Las Vegas

Troubleshooting :: Filesystem RAS events undeliverable

tmcnamara — Fri, 03 Sep 2010 07:06:01 GMT

Author: tmcnamara
Subject: Filesystem RAS events undeliverable
Posted: Fri Sep 03, 2010 2:06 am (GMT -5)

Seen this on two different xsans... system logs for the mdcs show the following when adding any new clients (here showing adding 192.168.1.101 w/ coordinators .10 and .11):

Aug 12 15:07:08 mdc1 servermgrd[73512]: Got error -9806 for SSLHandshake remote address is 192.168.1.101:4032
Aug 12 15:07:08 mdc1 servermgrd[73512]: Exception in threadListen: Socket: Connect failed
Aug 12 15:07:08 mdc1 fsmpm[115]: PortMapper: launching configuration reload thread
Aug 12 15:07:08 mdc1 fsmpm[115]: PortMapper: fsmpm configuration reload initiated (flags FFFFFFFF)
Aug 12 15:07:08 mdc1 fsmpm[115]: PortMapper: fsmpm configuration reload complete
Aug 12 15:07:08 mdc1 fsmpm[115]: Portmapper: File System RAS events undeliverable to Coordinator '192.168.1.11'. Please upgrade Xsan on this host.
Aug 12 15:07:09 mdc1 fsmpm[115]: Portmapper: File System RAS events undeliverable to Coordinator '192.168.1.10'. Please upgrade Xsan on this host.

Now Xsan is identical in all systems (v 2.2.1) and cvversions all reports 412.3; all systems are 10.6.4 and have separated static public/private networks, all with forward and reverse DNS (on both public AND private); all properly resolving and SAN is functioning normally. This happens across multiple systems added to the Xsan.

So this isn't causing any issues but is still something I'd like to know about; any reason this shows up in the log? I don't remember seeing this in 10.5 xsans on 2.1.

Troubleshooting :: RE: Finder crashes after xsan finder copy, Apple is stumped

smlsan — Thu, 02 Sep 2010 17:25:41 GMT

Author: smlsan
Posted: Thu Sep 02, 2010 12:25 pm (GMT -5)

We have reproduced the Finder hang with 4 modes: Icon Preview, Preview Column (column view), Quick Look and Get Info. All 4 of these appear to read through the file extents in order to generate the preview. Something about this process relates to the hang, although we are still researching the root cause. We can even get a hang to occur with Quick Look or Get Info after turning off Icon Preview and Preview Column.

Again we are only seeing this on 10.6 clients. We have tested 10.5 clients in parallel and had no hangs to date.

Check extent count on the file that hung: snfsdefrag -c <file>. If extent count is in the thousands this may be the same issue. When the hang occurs try doing 'ls -l' on the file in Terminal. The ls should hang as well. Sometimes the file that actually caused the hang is not the one you have trouble with. You may need to do a recursive ls on the directory tree to find the object that is actually hung.

You can defragment the file with snfsdefrag and should wind up with 1 extent which will eliminate the hang behavior for that file. For reasons discussed in other forum threads it isn't a good idea to run snfsdefrag on the whole filesystem or constantly, so use per file defrag and that is at best a temporary workaround.

We were able to eliminate the high extent count by increasing InodeExpand* parameters in the filesystem.cfg. InodeExpandInc is helpful, but InodeExpandMax was the main improvement. Defaults for XSAN 2.2.1 are:

InodeExpandMin 8 blocks (128K @ 16K block size)
InodeExpandInc 32 blocks (512K @ 16K block size)
InodeExpandMax 2048 blocks (32M @ 16K block size)

Filesystem needs to be stopped and restarted for those .cfg changes to take effect.

If you are working on very large files (100GB+) even the 32MB InodeExpandMax can lead to a few thousand extents under specific conditions. Check with Apple first, but it may make sense to increase those parameters beyond default. If they are set lower than default it may make sense to change them to default first.

Note, if you have only one process writing to a stripe group and the filesystem free space is not heavily fragmented you should get a file with 1 or maybe 2 extents. The high extent counts occur when writing two or more files at the same time in the same stripe group and they are competing for allocation of free space. That is where the InodeExpand* parameters kick in.

Please post a follow up with some of the extent count info. You can run a recursive 'snfsdefrag -c -r <dir>' for a directory tree, or use 'cvfsck -e' to get detailed extent information about every file in the filesystem. I'm very curious to see if you are running into the same issue we are working on.

Troubleshooting :: RE: Finder crashes after xsan finder copy, Apple is stumped

dmastroluca — Thu, 02 Sep 2010 16:47:05 GMT

Author: dmastroluca
Posted: Thu Sep 02, 2010 11:47 am (GMT -5)

Wow!! This looks like the same issue we are having. My Master Control Operator just tried to copy a large video file yesterday via finder and it triggered a hang. I sent stackshot logs and process samples for Finder and DirectoryService to our Apple engineer.

The Apple engineers told me to disable Icon Preview, but I had it disabled long ago thinking that was the problem. Since I had it disabled already, I did not see a difference in the frequency of hangs.

What were the tweaks you did to help with the number of extents created in a file?

Troubleshooting :: RE: XSan & Xserve RAID - Not Seeing LUNs

singlemalt — Thu, 02 Sep 2010 07:00:05 GMT

Author: singlemalt
Posted: Thu Sep 02, 2010 2:00 am (GMT -5)

You should also check how the HBA and or the RAID controller(s) is configured.
Should probably be Point-to-Point on one of them then automatic on the other. PtP on both should also work. And should probably be 2Gb for speed.

Following up on abstractude's question...
http://support.apple.com/kb/TS1268

Troubleshooting :: RE: XSan & Xserve RAID - Not Seeing LUNs

abstractrude — Wed, 01 Sep 2010 21:27:06 GMT

Author: abstractrude
Posted: Wed Sep 01, 2010 4:27 pm (GMT -5)

what do you get at the terminal with a
diskutil list

Troubleshooting :: XSan & Xserve RAID - Not Seeing LUNs

qrpike — Wed, 01 Sep 2010 16:54:52 GMT

Author: qrpike
Subject: XSan & Xserve RAID - Not Seeing LUNs
Posted: Wed Sep 01, 2010 11:54 am (GMT -5)

Okay, I have 2 peices of hardware, a Xserve (intel), and a Xserve RAID, 7x400gb and 7x750gb.

Software: OS X Server 10.6 and XSan 2.2

The raid is plugged directly into the xserve currently as there is not a need for a switch. I can see the RAID in the raid admin utility, I installed XSan and everything went smoothly. But now I want to actually add volumes but it does not see any volumes or LUNs.

Not sure what to do here, and there is very little info pertaining to this problem online. Any help would be GREATLY appreciated.

(fyi, raid is configured raid5).

Thanks,

Troubleshooting :: Make offline clips problem with xsan

Stavros — Wed, 01 Sep 2010 10:11:55 GMT

Author: Stavros
Subject: Make offline clips problem with xsan
Posted: Wed Sep 01, 2010 5:11 am (GMT -5)

Hi,
I have a problem with Final cut pro and Xsan.
Mac Pro with Leopard
Final cut pro 6
Xsan 2

We have 7 Mac pro and Xsan 2. When we make offline the clips in final cut pro and select delete from disk, the QT files are deleted from Xsan volume but in FCP in the browser does not shows the clips as offline (with the red line in the clips). Sometimes, when we leave FCP and click on the desktop and
come back to FCP, makes the clips offline, but this annoying
trick, does not always work.

Thank you.

Troubleshooting :: RE: Finder crashes after xsan finder copy, Apple is stumped

smlsan — Tue, 31 Aug 2010 21:44:16 GMT

Author: smlsan
Posted: Tue Aug 31, 2010 4:44 pm (GMT -5)

The issue we've run into is number of extents in a file, not number of extents in use overall. Issue occurs only on clients that access the file(s) that have very high extent count. Only occurs on 10.6 clients.

Disabling Icon Preview and Preview Column (column view) in Finder reduces the chances of a hang. We're doing testing now to see if copying a file in Finder can induce the hang. Note that move/rename will not because Finder does not read through the file for a move/rename and it is reading all the extent info that appears to trigger the hang.

We have recreated the problem in one of our labs with a new volume. Age of volume and amount of content does not have a direct impact, although a highly fragmented file system is more likely to create the right conditions that lead to the hang.

Repro case is fastest with a file >10,000 extents. Simply browsing a directory containing a file with that many extents will hang in column view with preview enabled. Touching the file (touch command in Terminal) will make Finder rescan the file and usually speeds up the hang occurrence.

I need to check with our client before sending their case number or any customer specific info.