Negative Seek Offset / VMFS FDC Base not found

[GoogleCodeExporter]

What steps will reproduce the problem?
1. java -jar fvmfs_r81.jar /dev/sda4 info

What is the expected output? What do you see instead?

should display /dev/sdb1 VMFS3 volume info

What version of the product are you using? On what operating system?

v0.9.8.14r. 81 / Knoppix v6.0

Please provide any additional information below.

Debugging Output Attached.

Original issue reported on code.google.com by trinkteu...@gmail.com on 3 May 2009 at 7:47

Attachments:

• debug.log

[GoogleCodeExporter]

Did you do something special when creating/formatting this FS, or was all done 
in a
"default fashion"? We need to be able to reproduce this issue.

Original comment by fluidops...@gmail.com on 9 May 2009 at 11:31

• Changed state: Accepted

[GoogleCodeExporter]

Issue 4 has been merged into this issue.

Original comment by fluidops...@gmail.com on 9 May 2009 at 11:34

[GoogleCodeExporter]

I also have this issue.  I just followed the wizard when creating the VMFS with 
VIC.
 the only difference between this and my other volumes is that this one is on SAS
disks and the others all sata, but all on the same SAS raid controller.

My debug dump seems to be much larger though, maybe it'll help

Original comment by cwe...@gmail.com on 12 May 2009 at 3:40

Attachments:

• debug.dump.bz2

[GoogleCodeExporter]

I have this issue too. Even on corrupted and working VMFS partition
All FSs were created in VIC
I'm attaching debug from corrupted partition

Original comment by emio...@gmail.com on 15 May 2009 at 11:11

Attachments:

• debug.zip

[GoogleCodeExporter]

so is it possible I also have a "corrupted and working" vmfs?  how could I tell?

Original comment by cwe...@gmail.com on 20 May 2009 at 3:06

[GoogleCodeExporter]

Hi there,
   I'm experiencing the same problem.

Our setup is as follows:

2 x Openfiler 2.3 SANs (SAN1 and SAN2) installed on Dell Power Edge Servers. 
Both
have 2x 1TB drives in RAID1 where we store the VM's in a volume which is 
exported to
ESXi v3.5 running on a third server.

Openfiler provides us with a read-only snapshot of the VM's every day at 00:00.

We have a single VM running Ubuntu 9.04 server (called BACKUP1) and Bacula 
software.
 The VM snapshot is mounted on to backup1 via a read-only ISCSI share.

I have successfully mounted VMFS volumes from both SANs and exported via webdav
(works great:-), but one particular VMFS volume causes the util to return the
"Negative seek Offset message".

I've looked through the log and compared it to your sample log, here's what 
i've noted:

1. We get an ignored "java.lang.NegativeArraySizeException"
2. We get an ignored " java.lang.ArithmeticException: / by zero" - nice!!!!

Is this any help?

Original comment by alex.sca...@googlemail.com on 20 May 2009 at 4:59

Attachments:

• log.txt

[GoogleCodeExporter]

hm, just now run into the same problem.
resolved it by observing a wrong partition flag - shoul be fb for VMFS
so try to go for fdisk /<device path> and than option under t chose fb.
w to write.
worked for me.

Original comment by pero.mud...@gmail.com on 20 May 2009 at 11:34

[GoogleCodeExporter]

mine is already fb
however, I'm pointing it at a LVM lv, I don't seem to have access to the 
partition
within, but this works on my other volumes, same layout.

Original comment by cwe...@gmail.com on 20 May 2009 at 11:51

[GoogleCodeExporter]

Unfortunately the team wasn't able to reproduce locally yet. Thanks to all so 
far for
testing & providing the logs -- trying to get a clue from the logs, otherwise 
may
request (part of) a FS dump to nail it down.

Original comment by fluidops...@gmail.com on 21 May 2009 at 2:15

[GoogleCodeExporter]

I think LVM might be at least part of the issue here.  I have LVM LV's exported 
via
scst-iscsi, then ESX puts a datastore on them.  my scst host doens't "see" the 
vmfs
partitions on the LV's, but fdisk does.  

to make it worse, I just make a new LV and formated it and added a guest to ESX,
fvmfs on the LV doens't see the guest dir I added, only the .fbb.sf, .fbc.sf,
.pbc.sf, .sbc.sf, and .vh.sf files and debug show no errors (which is also odd 
since
even the LUNS that work show some errors.)

So, if you put a vmfs on a disk, and them point fvmfs at the disk and not the
partition, what's that do for you?

Original comment by cwe...@gmail.com on 22 May 2009 at 5:03

[GoogleCodeExporter]

Has anyone come across any tools (from VMWare or otherwise) that can be used to 
scan
a VMFS file system and give it a health check? Or at least flag possible
issues/problems? Mush the same as fsck.ext does under linux? This would perhaps 
help
to identify if their is a problem with the VMFS filesystem and perhaps find a 
way in
which the FVMFS util can handle this gracefully.

You might argue that we shouldn't expect FVMFS to mount a file system thats 
damaged
in some way, and whilst I agree with this, the 9 VM's we have running on our 
ESXi
server don't appear to suffer any such problems if it is corrupted.

The mind boggles.......

Original comment by alex.sca...@googlemail.com on 27 May 2009 at 8:52

[GoogleCodeExporter]

[vmfs_r81]# /usr/java/jre1.6.0_13/bin/java -jar fvmfs.jar /dev/sda1 info
VMFSTools (C) by fluid Operations (v0.9.8.14 r81 / 2009-03-27_21-26-53)

java.io.IOException: Negative seek offset
Exception in thread "main" java.io.IOException: VMFS FDC base not found

Original comment by ke...@doublellama.net on 5 Jun 2009 at 4:31

Attachments:

• stack.txt

[GoogleCodeExporter]

Issue 18 has been merged into this issue.

Original comment by fluidops...@gmail.com on 16 Jun 2009 at 9:10

[GoogleCodeExporter]

I'm also having the problem, this is on an Windows 2003 Server, connected by 
Microsoft iSCSi to a Dell MD3000i.
On the MD3000i we have four VMFS volumes, 1.3TB each.
These drives can be seen in Windows (Disk 1 to 4, Disk 0 is the system disk).
When trying to query information on any of the drives, for example:
> java -jar fvmfs.jar \\.\PhysicalDrive4 info

it responds with:
VMFSTools (C) by fluid Operations (v0.9.8.14 r81 / 2009-03-27_21-26-53)
http://www.fluidops.com

java.io.IOException: Negative seek offset
[...]

Complete debug output is attached.

Original comment by alfschwa...@gmail.com on 17 Jun 2009 at 12:21

Attachments:

• debug.txt

[GoogleCodeExporter]

Original comment by fluidops...@gmail.com on 25 Sep 2009 at 6:16

• Changed state: Fixed

[GoogleCodeExporter]

We have a problem with a VMFS volume (see 
http://communities.vmware.com/message/1613557) and trying to troubleshoot the 
issue, we get this error:

root@muffin vmfs_r95# java -jar fvmfs.jar /dev/sda3 info 

Exception in thread "main" java.io.IOException: VMFS FDC base not found
at com.fluidops.tools.vmfs.VMFSDriver.openVmfs(VMFSDriver.java:1180)
at com.fluidops.tools.vmfs.VMFSTools.cli(VMFSTools.java:225)
at com.fluidops.tools.vmfs.VMFSTools.main(VMFSTools.java:492)

Results running FVMFS in debug mode are attached for a working datastore 
(storage5) and for the defect one (storage10). Any ideas what may be wrong? 
Thanks!

Original comment by xesc.arb...@gmail.com on 20 Sep 2010 at 4:46

Attachments:

• muffin_vmfs_debug_storage5.txt
• muffin_vmfs_debug_storage10.txt