No free block left after backup/image

[Gwendling]

Hy,

after a backup/image of a data disk, when i  mount it , he shows 0 blocks left.

a the of the backup , there is an error message that the quota file is too small (we don't use quota and the disks are mounted /noquota)

I need also to mount the disk /noshadow to be able to write on it (we don't use shadowing)

what can be the problèm and any idea to check/resolve.?

openvms alpha 7.2-3

Thank's in advance.

Gérard

[Steven Schweda]

> after a backup/image of a data disk, [...]

   What, exactly, did you do?  Copy+paste is your friend.

> [...] there is an error message [...]

   Copy+paste is your friend.

> [...] we don't use quota [...] we don't use shadowing [...]

   Are you sure that you're using the right disk(s)?  Are logical names
redirecting your commands to the wrong devices?

      HELP ANALYZE /DISK_STRUCTURE

[Gwendling]

Thank's  Steven,

here the answers :

[Dave Lennon]

That does seem strange.. A few questions/suggestions:

Is the original CAV1 disk part of a volume set? Can you include output from a SHOW DEV/FULL $1$DGA1:  ??

Were there any other error messages from the original BACKUP/IGNORE=INTERLOCK command?

Would it be possible to dismount/mount privately the $1$DGA1: source for the backup (therefore not needing the /IGNORE=INTERLOCK)? I see it had 3 open files on it, what were they? SHOW DEVICE/FILES $1$DGA1:

A defragmentation wasn't occurring on that disk while the bakup was going on, was it? $ SHOW SYSTEM/PROCESS=DFG$*

Everything is fine with $1$DGA1:, right? Files can be created there and an ANALYZE/DISK/REPAIR $1$DGA1: is clean?

[Gwendling]

Thank's Dave

Here the result of the sho dev

there is no DGF process in the system

Regards

Gérard

[Volker Halle]

Gérard,

please consider to copy & paste the screen output instead of posting screenshots. The major advantage of providing the text of the error messages is that those messages can be found using WEB search engines, which does not work with text in screenshots.

What does SHOW DEV/FULL $1$DGA166 look like, especially Total Blocks and Free Blocks ?

Consider to run ANALYZE/DISK $1$DGA1: first (without /REPAIR) and check for error messages.

$ HELP/MESSAGE QUOTAFILE tells you to rebuild the quota file (using $ SYSMAN DISKQUOTA REBUILD/DEVICE=$1$DGA166:). Did you try that ? Does that work ? If not, what are the error messages ?

Volker.

[Gwendling]

Thank's volker,

I'll test this next wednesday an post the result...

[Gwendling]

Hello

Here the answers :

What does SHOW DEV/FULL $1$DGA166 look like, especially Total Blocks and Free Blocks ?

$PROD>>sho dev $1$dga166:/ful

Disk $1$DGA166: (PROD), device type COMPELNT Compellent Vol, is online,

   allocated, deallocate on dismount, mounted, file-oriented device, shareable,

   available to cluster, error logging is enabled.

   Error count                   0   Operations completed           10339086

   Owner process           "SYSTEM"   Owner UIC                     [SYSTEM]

   Owner process ID       00002F3D   Dev Prot           S:RWPL,O:RWPL,G:R,W

   Reference count               2   Default buffer size                 512

   Current preferred CPU Id       0   Fastpath                             1

   WWID   01000010:6000-D310-059D-2C00-0000-0000-0000-00A6

   Total blocks           83886080   Sectors per track                   210

   Total cylinders           49933   Tracks per cylinder                   8

   Logical Volume Size     83886080   Expansion Size Limit         112852992

   Allocation class               1

   Volume label             "CAV11"   Relative volume number               0

   Cluster size                 96   Transaction count                     1

   Free blocks             14870688   Maximum files allowed           432402

   Extend quantity               5   Mount count                           1

   Mount status             Process   Cache name       "_PROD$DKA0:XQPCACHE"

   Extent cache size             64   Maximum blocks in extent cache 1487068

   File ID cache size           64   Blocks in extent cache               0

   Quota cache size               0   Maximum buffers in FCP cache       4160

   Volume owner UIC       [SYSTEM]   Vol Prot   S:RWCD,O:RWCD,G:RWCD,W:RWCD

Volume Status: ODS-2, subject to mount verification, file high-water marking,

     write-back caching enabled.

Consider to run ANALYZE/DISK $1$DGA1: first (without /REPAIR) and check for error messages

$PROD>>**bleep**/disk/repair/conf $1$dga1:

Analyze/Disk_Structure/Repair for _$1$DGA1: started on 14-OCT-2020 10:04:32.90

%ANALDISK-W-INCQUOTA, QUOTA.SYS indicates 31753818 blocks used,

       actual use is 39905495 blocks for [SYSTEM]

Repair this error? (Y or N): n

%ANALDISK-W-INCQUOTA, QUOTA.SYS indicates 1638 blocks used,

       actual use is 1474 blocks for [GIBAUDC]

Repair this error? (Y or N): n

%ANALDISK-W-INCQUOTA, QUOTA.SYS indicates 5900 blocks used,

       actual use is 31221 blocks for [GRANDJEANS]

Repair this error? (Y or N):

 HELP/MESSAGE QUOTAFILE tells you to rebuild the quota file (using $ SYSMAN DISKQUOTA REBUILD/DEVICE=$1$DGA166:). Did you try that ? Does that work ? If not, what are the error messages ?

gives a "no such file error ( i deleted the quota.sys file as we don't ue quota

THe problem may du to the errors on dga1:  I dint't do th repair, we will use a backup command without image.

Thank's to all

Gérard

[Volker Halle · Accepted solution]

Gérard,

$1$DGA1 and $1$DGA166: have the same number of total blocks.. But the 'cluster size' is different - why ? If you would have lots of small files they will be allocated using the bigger cluster size on $1$DGA166: and will consume more space on that disk. BACKUP/IMAGE should normally init the destination disk with the same parameters as the source disk.

If you don't use diskquotas, consider to disable the use of DISKQUOTA on $1$DGA1 first and then delete QUOTA.SYS on $1$DGA1:

Using BACKUP without /IMAGE may create other problems...

Volker.

[Gwendling]

Volker

I didn't notice the différence in cluster size...we initialised $1$DGA166: with no parameters, we then have done this backup:

bac/ign=inter $1$dga1:[*...]*.*;* $1$dga166:[*...]*.*;*/by_owner

this was just a test transfert for the real one i will init the destination disk with the same cluster size.

Many thank's

Gérard

[Volker Halle]

Gérard,

in your first screenshots, you've used BACKUP/IMAGE, that should have correctly initialized the output disk. Were these screenshots from different tests ?

Volker.

[Gwendling]

Yes Volker that's true

i used first /image and faced the 0 free block left problem. i then used the other backup.

[abrsvc]

I would have expected the BACKUP/IMAGE to initialize the disk with the same clustersize as the input disk.  You can either initialize the disk prior to the backup and specify BACKUP/NOINIT or just make sure that there is no symbol set up for backup that is turning off the initialization stage.  As you have seen, the difference in clustersize will cause the space usage change you are seeing.

Dan

[Gwendling]

Thank's Dan i Will use a backup noimage and noinit alter manual initilization.

regards

Gerard

[abrsvc]

No.  Please use BACKUP/IMAGE for the entire disk.  The key here is to add the qualifier /NOINIT.  This will make sure that your initializaton won't be overwritten.  Truthfully, BACKUP/IMAGE should have preserved the clustersize of the disk being copied, but for whatever reason it did not.  An IMAGE copy is required here to preserve any back links etc.

Dan

[Gwendling]

Hello Dan,


As it is not a system disk i think i can use a backup of [*...]*.*;* with the /by_owner qualifier on the destination.


I think the backup/image on the first post did the initialisation with the same clustersize as the source disk. The différent clustersize later in this post is the result of a manual initialization and backup. ( result of différent trys)


Btw i will try the /image/noinit ...


Thank you fit your help

Gerard