apgl's picture

Fatal Metadata inode error

Dear friends:

My Xsan system did not mount after a Metacontroller transfer. When using cvfsck -wv command, the following error was given:

Attempting to acquire arbitration block... successful.

  • Fatal*: Metadata inode for stripe group 'MetadataAndJournal' has been damaged. Giving up.

PANIC: cvfsck ASSERT failed "idinobuf.idi_flags & InodeFlagMetaData" file cvfsck.c, line 5832
Abort trap

Appreciate someone could help me fix this.

Hi Refinition's picture

X serve locks up at log in

We are running a G5 Xserve (running 10.6) as are Primary MDC for a Xsan. I noticed it was locked up about a week ago and that the secondary MDC was running the Xsan. When we rebooted the Xserve it will go to the log in screen and after about 2 seconds we the rainbow wheel. I have tried booting up resetting the NVRAM, in Verbose mode, in Safe mode, in Single user mode and I even tried booting off the install disc too but all lead to the rainbow wheel lock up. We pulled the fiber card (even switched it with another), reseed the RAM but still nothing works.

Last resort is to rebuild and reformat... which I really do not what to do since I am not a IT guy but just have enough knowledge to really screw things up.

Any ideas or something I have not tried?

Gerard's picture

Xsan 2.2 Permissions Issue

16Tb Xsan
Xsan 2.2
Two fiber attached stations
Non fiber users connecting via nfs auto-mount
2 MDCs
Using ACL to set permissions.

Experiencing permission issues when users create, read, write files. It is very sporadic of a case. Some users will be able to read/write
files/folders & there are some who cannot with the same files/folders.

For example:
One user will create a folder and add in content, but others users will have trouble accessing the data.

Propagating permissions, via Xsan Admin, usually fixes the issues, but it doesn't resolve the issue of why this is happening.

kjaer.casper's picture

Promise X30 issue - Command times out on 10 disk


Hi all.

I get theese errors on my X30 48TB raid , 10 disks makes command time out at same time, all 10 disk, in that raid6 is controlled by controller 2. Promise has sent me a new controller, and i have installed it, but the command times out is still here.

When i changes all traffic to controller 1, i dont see the issue. I have now received a new chassis from Promise, that we are gonna try.

The command time out, makes XSAN crash totally, finder is not responding on all machines, including metadata controllers. Which is normal i guess , if XSAN is loosing a LUN.

Any suggestions???


Error log from NVRAM events.

417PD 240x000D0003InfoNov 4, 2012 21:51:09Physical Disk has been reset416PD 240x000D0011MinorNov 4,
2012 21:51:09Command times out on physical disk415PD 230x000D0003InfoNov 4,
2012 21:51:09Physical Disk has been reset414PD 230x000D0011MinorNov 4,
2012 21:51:09Command times out on physical disk413PD 200x000D0003InfoNov 4,
2012 21:51:09Physical Disk has been reset412PD 200x000D0011MinorNov 4,
2012 21:51:09Command times out on physical disk411PD 190x000D0003InfoNov 4,
2012 21:51:09Physical Disk has been reset410PD 190x000D0011MinorNov 4,
2012 21:51:09Command times out on physical disk409PD 160x000D0003InfoNov 4,
2012 21:51:09Physical Disk has been reset408PD 160x000D0011MinorNov 4, 2012 21:51:09Command times out on physical disk407PD 120x000D0003InfoNov 4,
2012 21:51:09Physical Disk has been reset406PD 120x000D0011MinorNov 4, 2012 21:51:09Command times out on physical disk405PD 110x000D0003InfoNov 4,
2012 21:51:09Physical Disk has been reset404PD 110x000D0011MinorNov 4, 2012 21:51:09Command times out on physical disk403PD 80x000D0003InfoNov 4,
2012 21:51:09Physical Disk has been reset402PD 80x000D0011MinorNov 4, 2012 21:51:09Command times out on physical disk401PD 70x000D0003InfoNov 4,
2012 21:51:09Physical Disk has been reset400PD 70x000D0011MinorNov 4, 2012 21:51:09Command times out on physical disk399PD 40x000D0003InfoNov 4,
2012 21:51:09Physical Disk has been reset398PD 40x000D0011MinorNov 4, 2012 21:51:09Command times out on physical disk397PD 150x000D0003InfoNov 4,
2012 21:07:14Physical Disk has been reset396PD 150x000D0011MinorNov 4, 2012 21:07:14Command times out on physical disk

morphenine's picture

Premiere Pro CS6 Permissions

We're evaluating moving to Premiere Pro CS6 from FCP7. We ran into a problem and I'm wondering if anyone else has hit this wall...

Basically if edit station A saves a project, then edit station B can open and manipulate the project but cannot save it unless he is the POSIX owner of the file. Premiere Pro doesn't seem to respect Group or Other permissions, nor the ACLs.

Going to reach out to Adobe tomorrow.

xsanguy's picture

ACLs are busted - force some clients to mount a volume RO?

ACLs aren't working on a fairly large production XSan. The auto-complete in group-name when adding an ACL takes 20+ hours (!) This is a large system, a couple Petabytes and millions of files.

There are 5 Volumes, we would like to mount 2 of them RO on some of the client machines, the rest RW.

Is there a way to force certain clients to mount RO?

Thawk9455's picture

Meta Data Lun Suggestions

I'm curious what others see as typical I/0s per second and what benefits we could expect from possible changes to our metadata luns. We are using our Xsan as general file server storage so there is a wide array of data types from large numbers of small files all the way up to 200+GB files depending on the user. We're currently using about 80 TB of storage and recently have started seeing much slower overall performance when accessing the file server.

According to the Promise Performance Monitoring our Meta Data Lun is regularly in the 200-300 I/Os per second.

We are on our original setup, with two 750 GB SATA drives mirrored and are considering a few options but don't have the money to evaluate all of them individually so I'm hoping others that have worked with these configurations can provide some guidance.

1. We're curious if anyone has replaced their metadata drives with SSDs. I'm assuming this would provide a significant performance increase but also not sure what drives would work with the Promise E610s (Apple version) and offer the reliability we obviously would want for this type of system.

2. Replace the SATA drives with higher speed SAS drives. I know we can get 1 TB 10K SAS drives that could just be swapped out with the spare and migrated one at a time. How much of a boost could we really expect from a slightly faster drive?

3. Try to beg and plead our way into getting a new x30 and migrate the metadata drives to a RAID 10 configuration with more drives involved. Seems like kind a waste to burn up drives but if it will significantly improve overall read/write performance it could be worth it.

Thoughts? Suggestions? Are we missing something simple?

dshott's picture

Replacement Apple Xraid Drives


Hi, I was wondering if anyone has any suggestions on a place to buy replacement drives (750GB) for the old Apple Xraids? Thanks!

aaron's picture

Xsanity is Back!

As you guessed, the hurricane knocked out our multiply-redundant data center. (OK, it's really an Xserve G4. But it used to be awesome.)

The power is back on 23rd Street. Thanks to Con Edison for working so hard, and to all the Tekserve employees for working by candlelight for these last few days. The energy poured into perseverance has been very impressive, but it's good to have light!

Kasper Winding's picture

Qnap NAS + Atto Xtend issci software + Xsan 3.0 problems

Hi all

I got a problem with Qnap NAS + Atto Xtend software iscsi + Xsan 3.0

The issue is that the Qnab NAS (TS-879U RP) with firmware 3.7 is not showning the serial numbers on the HDD's. And that causes Xsan to not being able to detect multiple LUN's. Actually looks like Xsan detects them randomly and always one at a time.

Is there any firmware addons, 3. part etc. that can solve this HUGE problem?.

Best Regards
Kasper Winding


Subscribe to Xsanity RSS