Device Node Change time to time
Hi,
I have a small question to you all.
I have seen this now multiple times and can't really explain why it's happening.
If I start with my hardware I got.
I have a couple of IBM x3650m2 and HP380G5 but all of them have multiple Qlogic 24xx 4GBit FC controllers that is connected to a Broadcade FC Switch, as back-end storage do I have a IBM DS-Storage Disk approx 500TB with RAID6.
Anyway, I have create LUNs on 10TB each.
I have a mix of Linux Dist with RHEL 5U3 and SLES11
After installation OS on local harddisks and configure MPIO and so on to my IBM DS-Storage.
I start with PARTED and also tried FDISK but both react the same way.
When I'm finish with the partition and go to /dev/mapper/, then can I see my Device Node called for ex. disk0p1 (10TB partition) and I can format that with EXT3 (Company policy) and I reboot my server and I can still see disk0p1 as device node.
But after a couple of days and maybe a couple of reboots later does disk0p1 disappear and I can only see disk0.
If I run "mount /dev/mapper/disk0 /mnt/" then can I still read my partition as normal and still read/write to that device without any problem and dmesg or messages log doesn't show any errors.
Funny part with this issue is that it doesn't happening on all device nodes. At the moment do I have following layout one of my servers.
/dev/mapper/disk0
/dev/mapper/disk1
/dev/mapper/disk2
/dev/mapper/disk3p1
/dev/mapper/disk4
/dev/mapper/disk5p1
If I'm running parted or fdisk to any partition that end with p1, fdisk or parted doesn't show any partition information and tells me that disk is empty.
Why is this happening? and how can solve this?
Cheers
Chris
|