parameters for cloning/init the system disk ?

[Christian Burschka]

Dear experts,
once more I would like to learn from Your wisdom...

I am about to clone the system disk dka0: on an
alphastation 255 with vms 6.2-1H3

Being not familiar with this task I booted from CDrom, initialized the second disk with no parameters, then after
"mount/foreign dka300:
did
"backup/ima/veri dka0: dka300: /noinit

there was a message that indexf.sys had to be extended (?) and that 3 files,
pagefile.sys , swapfile.sys and (cant remember ??) were marked "nobackup" and not copied.

then
dismount dka300:
mount dka300: ...
show dev /full dka300: and the same with dka0:

(now I wonder if these numbers make sense:)

dka0: blocks 2050860 clustersize 3 ext.quant 3
maxfiles 256357
maximum blocks in extent cache 1771

dka300: blocks 8388315 clustsize 5 extquant 5
maxfiles 419415
maximum blocks in extent cache 633046

my question is:
What should I do to get a nicely working system?
I imagine I have to set up a new pagefile.sys, swapfile.sys and (??).
run sysgen/autogen ? what else ?
or redo the backup ?
may be with more appropriate/suitable disk parameters ?

Although I would not recommend it, I would like to use unoccupied space for data storage on the new system disk dka300:.

Any suggestions gratefully appreciated.

Regards
Chris.

[Volker Halle]

Chris,

you should be able to boot the copy of your system disk from DKA300. No need to run AUTOGEN, if you have just changed the physical system disk. You may still want to run @AUTOGEN SAVPARAMS SETPARAMS FEEDBACK after a couple of days.

BACKUP did not copy the contents of page, swap and dump files (these files are marked /NOBACKUP), as this contents is meaningless on the copy of the system disk.

Volker.

[Karl Rohwedder]

Regarding data storage on the system disk:
If you cannot avoid it due to number of disks available, I would suggest using the logical disk driver (LD found at http://www.digiater.nl/). It simulates a real disk but hard limits the number of blocks your data may use. This avoid situations, where the system disk is filled up until the AUDIT_SERVER blocks all other activity.

regards Kalle

[comarow]

It is normal for the pagefile and swapfile to be marked "no backup". Thus, you simply create empty files with no data on it on your target disk. The data on those disks is only for temporary use, so it should not be backed up, much less restored. It will however create a contiguos file.

You made a mistake doing the backup with the
/noinit qualifier. It based the size of the indexf.sys to default values and they are too small. If you got rid of the /noinit qualifier, it would create a larger indexf.sys file.

That said, there are many situations to use /noinit. The main ones being if you want to INCREASE the size of the indexf.sys. thus to pre-allocate it for additional use later, and if you want to change the cluster or other such settings. Since you did none of these, your disk will be cleaner with simply

$Backup/image

Have fun.

[Jan van den Ende]

Chris,

_IF_ you can manage to do the exercise again, I suggest you begin with INITIALISEing the target disk with /CLUSTER=4
Do a SHOW DEVICE/FULL, and compare that with HELP INIT for various other settings.
Some other good ideas would be /NOHIGHWATERMARKING, and /EXTEND=, although those can also be changed later.

Do not forget /OWN=SYSTEM (or /OWN=[1,1]; your choice).

After that, do the BACKUP/IMAGE again, but _WITH_ /NOINIT.
The whole idea about /CLUSTER=4 (or 8 on bigger drives) is to make the disk allocation factoring nicely into the chunks of data tranfer and of memory allocation, which are also multiples of 8.

Your gain in performance will be worth your efforts!

Success.

Proost.

Have one on me.

jpe

[Christian Burschka]

Dear experts,

thank You all for Your comments.

I first did the init without parameters,
relying on the intelligence of the system.
(Is the system intelligent , or does it simply use certain defaults ?)
Then did the backup with /noinit specified,
to use the system generated parameters of the disk, partially ok ?

I will do the init and backup (/noinit) again:

init with clustersize=4
/extend=256 (instead of 5)
/nohighwater

does this mean that the disk should get less fragmented ?

ok that pagefile.sys etc. are not "copied" by backup. I assume that it is at least created
with about the same (?) size ? Do I have to adjust their size on the new disk ?

Regards,

Chris

[Christian Burschka]

comarow,

please, how should I calculate the suitable
size for indexf.sys ? How should I proceed about it ?

I understand it that backup just takes it from the old system disk, no ?

Regards,

Chris.

[Karl Rohwedder]

Just estimate, how many files will be created on the disk and specify this value to /HEADERS=nn (see HELP INIT /HEADER).

regsrda Kalle

[Robert Gezelter]

Chris,

The PAGEFILE and SWAPFILE will be the same size as the original files. As has been noted, they have no meaning outside of a running system, so they are marked NOBACKUP.

The EXTEND size is a judgement call, depending on the activity on the disk. For normal, small text files, 5 is not a particularly bad value. If you are writing BACKUP Save Sets to the volume, 10,000 is a more realistic number. In any event, the volume extend size is only the default, the extend size is controllable on a process basis (with SET RMS) and on a per-file basis (directly through RMS and the various RTL routines).

- Bob Gezelter, http://www.rlgsc.com

[Volker Halle · Accepted solution]

Chris,

may I just suggest, that you just perform a BACKUP/IMAGE DKA0: DKA300: and boot from your new copy of the system disk.

Trying to 'optimize' the disk parameters without being able to actually measure any possible improvements (in the milli-seconds time domain) may just be a waste of time.

Volker.

[Christian Burschka]

Volker,

thanks,

to hear this from an expert, gives me much relief....

Regards

Chris.

[Christian Burschka]

Thanks again to all of You

for Your valuable comments.

Regards Chris