This is a weird problem that has been driving me crazy - so please help!

I have a Linux box with external scsi storage.

Red Hat Enterprise Linux AS release 4 (Nahant Update 4)
Linux njserver1 2.6.9-42.ELsmp #1 SMP Wed Jul 12 23:27:17 EDT 2006 i686 i686 i386 GNU/Linux

The external SCSI drive is attached to the system using Adaptec PCI SCSI card. Here are some of the details regarding PCI SCSI card and the hard disk itself.

Adaptec AIC7XXX EISA/VLB/PCI SCSI HBA DRIVER, Rev 6.2.36
<Adaptec 2940 Ultra SCSI adapter>
aic7880: Ultra Wide Channel A, SCSI Id=7, 16/253 SCBs
Vendor: HITACHI Model: HUS151436VL3600 Rev: S3C0

Problem is that the extended partitions that I create on the external SCSI disk (using fdisk) disappear after some time. The 'fdisk -l' command doesn't show the extended partitions. However, the data stored in those extended partitions continue to be available. Is this a partition table corruption? I have to reformat and re-partition the hard disk to fix this issue. However, after few days, the issue re-appears.

Has anyone seen an issue like this. Any feedback will be highly appreciated!

If it's true SCSI it shouldn't need extended partitions. SCSI has up to 8 primary partitions, if this isn't confirmed by your fdisk I'd double check that it's SCSI, or the correct driver is being used for it (although, I can't imagine the wrong driver being used).

Regards,

Alunduil

Thanks, alunduil.

However, I am not sure what is meant by True SCSI. I have the Adaptec Ultrawide 2940 SCSI PCI card and Hitachi 36.7 GB SCSI harddisk housed in an external casing.

Also, in this case, maximum of 4 primary partitions is a limitation of Linux Kernel. The version I have won't let me create more than 4 primary paritions.

The extended partitions stay there for few days and then disappear. Maybe partition table is getting corrupt. Will installing an older aic7xxx driver help?

Depending on whether or not your kernel is using libATA drivers, the kernel won't see anything above the #15 partition... a SCSI limit.

I only had 9 partitions total - 3 Primary and 6 extended. Now, I have only three left.

Disk /dev/sdb: 36.7 GB, 36703949824 bytes
255 heads, 63 sectors/track, 4462 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/sdb1 1 499 4008186 83 Linux
/dev/sdb2 500 998 4008217+ 83 Linux
/dev/sdb3 999 1497 4008217+ 83 Linux
/dev/sdb4 1498 4462 23816362+ 5 Extended

Here /dev/sdb5 thru sdb9 are gone.

i know this doesn't help much, but if you recreate the partitions with exactly the same starting and ending cylinders as before then the data should still be there when you reboot.

maybe change the extended type 05 to 85, linux extended, just to do something different.

I believe you mean logical not extended partition. If this is not a typo please explain how you are creating these "extended" partitions.

An extended partition is in a nutshell is a container for logical partitions so one can create more then the orginal 4 primary partitions. PC drives get partitioned the same way no matter if they are IDE, SCSI, SATA etc.

Interesting, I've never had partitions just disappear. If the partition table was corrupted then I would not think the data would continue to be available.

O.K., so you're really talking about logicals - I was also thrown by the reference to "extended".
I would expect things to keep working until the next boot - /proc/partitions would see to that.
Go find testdisk if you don't already have it- it'll recover the partition table for you, save you having to do the reformat you're doing.

I'd be looking for another driver to test.

Sorry if I used incorrect/confusing terminology while describing the problem.

1) I am creating 3 Primary partitions (/dev/sdb1 thru /dev/sdb3)
2) The fourth one I create as an extended partition.
3) Then, I create logical partitions within that extended "container" partition.
(/dev/sdb5 through /dev/sdb10)

I tried whansard's suggestion and created the extended partition as type "85" (Linux extended) instead of the default type "5" (Extended). I would keep monitoring and see how long these last.

Here is a "fdisk -l" output:

Disk /dev/sdb: 36.7 GB, 36703949824 bytes
255 heads, 63 sectors/track, 4462 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/sdb1 1 499 4008186 83 Linux
/dev/sdb2 500 998 4008217+ 83 Linux
/dev/sdb3 999 1497 4008217+ 83 Linux
/dev/sdb4 1498 4462 23816362+ 85 Linux extended
/dev/sdb5 1498 1996 4008186 83 Linux
/dev/sdb6 1997 2495 4008186 83 Linux
/dev/sdb7 2496 2994 4008186 83 Linux
/dev/sdb8 2995 3493 4008186 83 Linux
/dev/sdb9 3494 3992 4008186 83 Linux
/dev/sdb10 3993 4462 3775243+ 83 Linux

syg00:

The partitions above survive across reboots. They are just disappearing after few days. Are we saying that "logical" partitions within "container" (extended) partition are supposed to last until the next reboot?

Thanks everyone for your responses. Appreciate it.

All partitions should stay put - forever, or until you do something to delete them.
No question. As above, I am also unable to explain what the hell is happening. I have never before heard of such a thing.

My comment was in regard to the partitions "disappearing" but you still be able to reference the data. That would only be viable until the next boot - after which the data would be unreachable.