On 10/19/10 9:17 AM, Joe Balenzano wrote:
Alex,
From what I understand it looks like his vdisk gets a new VTOC label since he
removed the disk and added it back. I would expect this.
I just want to confirm a few things.
Yes, here the vdisk backend (a file) was removed and a new, larger backend
file was
created.
- He could of simply just unbound the ldom, grow the disk image file, then bind
the the ldom.
Nno need to remove and add it back? If he had done this, his partition table
would not have
changed (other than the size of slice 2 with S10 U8+ or withn patch 139562-02 )
In that case, the partition table would not have changed at all, and that's
currently
a problem because there's currently no easy way to adjust an existing partition
table
to a new disk size (other than some format(1m) tricks).
He would still have to run format to repartition it (giving his slice 0 all of
the disk).
- If he originally had exported the disk with the slice option, there would be
no need to run format in the guest, correct?
In his case, this would of been ok because he dedicating the entire disk to a
single partition.
Yes, with the slice option, the partition table and slice 0 is automatically
adjusted
to the size of the backend.
alex.
Alexandre Chartre wrote:
You first need to run format -> label (or format -> partition -> label) before
you can use any partition from the vdisk. By default, vdisks do not have a
partitioning. The number of cyl/hd/sec looks good:
(1846 + 2) * 96 * 768 * 512 = 69,759,664,128 = 64.97 GB
alex.
On 10/19/10 7:01 AM, Andrew Williamson - Fujitsu wrote:
Hi
Can anyone advise what's gone wrong here please ?
I had a test machine with a 60g backend (data.img below) on one of the disks. I needed it
to be 65g and the data in it can be re-cloned from the live system, so I thought,
"not a problem, stop the domain, remove the device, create a new backend, start the
domain, newfs, re-clone, be happy".
Fail.
It appears the LDOM is seeing the disk wrongly, check the parameters that
format reports below:
AVAILABLE DISK SELECTIONS:
0. c0d0<SUN-DiskImage-8GB cyl 27960 alt 2 hd 1 sec 600>
/virtual-devi...@100/channel-devi...@200/d...@0
1. c0d1<SUN-DiskImage-4GB cyl 13979 alt 2 hd 1 sec 600>
/virtual-devi...@100/channel-devi...@200/d...@1
2. c0d2<SUN-DiskImage-100MB cyl 339 alt 2 hd 1 sec 600>
/virtual-devi...@100/channel-devi...@200/d...@2
3. c0d3<SUN-DiskImage-1GB cyl 3493 alt 2 hd 1 sec 600>
/virtual-devi...@100/channel-devi...@200/d...@3
4. c0d4<SUN-DiskImage-10GB cyl 34950 alt 2 hd 1 sec 600>
/virtual-devi...@100/channel-devi...@200/d...@4
5. c0d5<SUN-DiskImage-65GB cyl 1846 alt 2 hd 96 sec 768>
/virtual-devi...@100/channel-devi...@200/d...@5
96 heads?? 768 sectors?? Cylinders is way too few!
6. c0d6<SUN-DiskImage-1GB cyl 3493 alt 2 hd 1 sec 600>
/virtual-devi...@100/channel-devi...@200/d...@6
7. c0d7<SUN-DiskImage-1GB cyl 3493 alt 2 hd 1 sec 600>
/virtual-devi...@100/channel-devi...@200/d...@7
8. c0d8<SUN-DiskImage-4GB cyl 13979 alt 2 hd 1 sec 600>
/virtual-devi...@100/channel-devi...@200/d...@8
Disk 5 is way wrong. Not even remotely working.
# newfs /dev/rdsk/c0d5s0
/dev/rdsk/c0d5s0: I/O error
I don't have a copy of what it should look like before the resize. The question
is, why has this happened, and what to do about it ?
Here's what I did to re-generate my larger file.
Before starting:
ldm ls -l primary
vdisk0 train-dom-r...@primary-vds0 0 d...@0 primary
vdisk1 train-dom-s...@primary-vds0 1 d...@1 primary
vdisk2 train-dom-us...@primary-vds0 2 d...@2 primary
vdisk3 train-dom-worksp...@primary-vds0 3 d...@3 primary
vdisk4 train-dom-ora...@primary-vds0 4 d...@4 primary
* vdisk5 train-dom-d...@primary-vds0 5 d...@5 primary
vdisk6 train-dom-l...@primary-vds0 6 d...@6 primary
vdisk7 train-dom-sp...@primary-vds0 7 d...@7 primary
vdisk8 train-dom-fi...@primary-vds0 8 d...@8 primary
ldm ls -l houtrain
NAME VOLUME OPTIONS MPGROUP DEVICE
primary-vds0 train-dom-root /ldoms/houtrain/sys/root.img
train-dom-swap /ldoms/houtrain/sys/swap.img
train-dom-users /ldoms/houtrain/sys/users.img
train-dom-workspace /ldoms/houtrain/sys/workspace.img
train-dom-oracle /ldoms/houtrain/tpp/oracle.img
* train-dom-data /ldoms/houtrain/dpp/data.img
train-dom-logs /ldoms/houtrain/dpp/logs.img
train-dom-spool /ldoms/houtrain/app/spool.img
train-dom-first /ldoms/houtrain/app/first.img
bash-3.00# ldm remove-vdisk vdisk5 houtrain
bash-3.00# ldm remove-vdiskserverdevice train-dom-d...@primary-vds0
bash-3.00# rm data.img
bash-3.00# mkfile 65G data.img
bash-3.00# ldm add-vdiskserverdevice /ldoms/houtrain/dpp/data.img
train-dom-d...@primary-vds0
bash-3.00# ldm add-vdisk vdisk5 train-dom-d...@primary-vds0 houtrain
Started the domain, since then, I/O errors as above, system drops into single-user coz it
can't fsckall. That's not a problem in itself, I'm just wondering how come the system
things this "new" disk is so different to all the others ??
/dev/rdsk/c0d2s0 is clean
/dev/rdsk/c0d3s0 is clean
/dev/rdsk/c0d4s0 is clean
/dev/rdsk/c0d7s0 is clean
/dev/rdsk/c0d8s0 is clean
checking ufs filesystems
Can't open /dev/rdsk/c0d5s0: I/O error
/dev/rdsk/c0d5s0: CAN'T CHECK FILE SYSTEM.
/dev/rdsk/c0d5s0: UNEXPECTED INCONSISTENCY; RUN fsck MANUALLY.
fsckall failed with exit code 1.
How to fix this, but more importantly what went wrong ??
xxx_sc> showhost
Sun System Firmware 7.2.10 2010/07/19 17:13
Host flash versions:
Hypervisor 1.7.9 2010/07/19 15:51
OBP 4.30.9 2010/07/16 09:01
POST 4.30.9 2010/07/16 09:39
Thanks
AW
______________________________________________
This email has been scanned by Netintelligence
http://www.netintelligence.com/email
_______________________________________________
ldoms-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/ldoms-discuss
_______________________________________________
ldoms-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/ldoms-discuss
_______________________________________________
ldoms-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/ldoms-discuss
