vms disk help

[Matthew Murdock]

Upon booting vms, i get quue manager cannot start...needs 3522 free blocks....So I boot into (what unix calls) single user mode on alpha.
>>>boot -FLAGS 0,1
SYSBOOT>set/startup opa0:
SYSBOOT>set window_system 0
SYSBOOT>set writesysparams 0
SYSBOOT>continue (does its thing)
$>define/system/exec decw$ignore_workstation true
$>spawn
$>mount/override=id dka0:
dka0:[SYS0.SYSCOMMON.][SYSLIB]MOUNTSHR.EXE;1
Protected images must be installed
$>
Now How do I mount the all the VMS partitions (or volumes)so I can find some files to delete or logs to delete? I need 3522 Blocks free for queue manager to start during normal boot.

[marsh_1]

hi,

if you do a

$ sh dev d

you should find the system disk already mounted, this where the queue manager database and journal files would be, the display from sh dev d will give you the free space on the volume.

hth

[Robert Gezelter]

Matthew,

There are a couple of ways of doing this. One is to do a Minimum boot (SET STARTUP_P1 "MIN").

When I want to gain control to fix something, I often use the SET /STARTUP=OPA0: trick. In that case, the system is not fully up when the "$" prompt is reached. Generally, I do a SPAWN, and then execute STARTUP.COM.

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

[David Jones_21]

Instead of 'set/startup opa0:', try 'set startup_p1 "MIN"' as your first sysboot comand. This will do a minimum startup.

[Matthew Murdock]

Mark > Ok I saw that it was mounted, but what do i do next?

Robert, David > Ok I did that then the second sysboot> I typed in "continue", it continued to try to start Decnet (which isnt licensed), the at the $, my keyboard stopped responding. In a previous attempt yesterday, I beleive I typed in set noon.

[Robert Gezelter]

Matthew,

The attempted startup of DECnet should not be a problem. For that matter, there is really no need to reboot in the first place.

The loss of available space is most likely caused by some files being continuously extended and not archived. Log files are common culprits, but I have seen many other causes.

Correct the space problem, and the Queue Manager can be restarted without a reboot (This is OpenVMS).

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

[marsh_1]

hi,

try

$ dir/siz/sel=siz=(min='number_of_blocks') sys$sysdevice:[000000...]*.*;*

to help you narrow down the search, do you know what you are looking for ?

fwiw

[Hoff · Accepted solution]

OpenVMS Tips: Your System Disk is Full? DEVICEFULL? ...
http://labs.hoffmanlabs.com/node/939

[comarow]

One additional step. Sometimes files are
deleted when they are open so they are simply
marked for delete.

When the system is quiet do an
Analyze/disk/repair to free the deleted files
(and find other problems on the disk.

BTW, some files get huge over time
such as the errorlog file, accountng.dat
$set account/new will create a new accountng.dat file.
mcr jbcsyscommand 1
will help clean up your queue file.

Store those under steps to take.

Have fun,
Bob

[Hoff]

FWIW...

If you can't find two megabytes of file storage to nuke, you're probably way too close to your installed disk capacity for comfort and best operation.

Operating close to capacity has its direct problems and direct consequences (such as this case) and also tends to cause indirect problems such as excessive disk fragmentation and the associated and lowered I/O performance, and hotter drives.

There are OEM 300 GB SCSI disks around (new) for under US$350 (some assembly required), and thus replacing too-small spindles is often a viable an option. On the used-equipment auction market, the DEC and Compaq 18 GB StorageWorks drives are about the bottom tier disks available when last I looked; anything smaller isn't usually worth buying (barring cases when the equipment vendor provides a warranty, if you want that); for geographies with sufficient installed hardware bases, 18 GB and smaller tends to be trash, or "haul it off for free" size.

[comarow]

I'd really like to point out that about the worst thing to do with a filled disk is reboot,
as rebooting is very difficult. Unless the
accounting file is the culprit you can
do some emergency stuff on your system disk
to get it working.

Purge/keep=3/log sysdisk:[*...]
will let you see what's writing a lot of files to the system.

new operator log, errorlog.sys, and accountng.dat files will help. As long as they are backed up you can always go back to them or save them on another disk.

A note, if you do a
set account/new it will start a new file
BUT IT WILL INCLUDE all Image activation,
so immediately do a set account/noimage

Since these files are created a bit at a time,
they may take up 100s of retrival pointers,
and the lack of free headers could stop you in your tracks. If you want to keep the existing files, such as accountng.dat
do a copy/contiguous filename Then purge the
original. So you have some options.
You can use the convert and recreate the file
with extra space.

All the log files have this problem. I liked
to use the RMS utilities to convert them to pre-sized files with extra space. Remember that keeping all your fragmenting log, or dat
files will have all these pointers. This also means if you use the files, such as use accounting.dat or your system backs it up it requires many extra I/Os. So this has performance issues.

Then, do the dir/sel= look for big files
and get huge files that may be sitting wasting space.

My rule of thumb is keep no applications on the system disk.

Finally, over the years, even sysuaf.dat
can, like any other ISAM files get it terrible shape with internal bucket splits,
records marked deleted. Since this file is hit in every login, including batch, it's worth optimizing it.

For example

Anal/rms/fdl sysuaf.dat
The edit/fdl syauaf.fdl
and choose optimize. It will create
an FDL file for your sysuaf.dat.
Make sure you choose the option to
add additional records, and the fill factor low and it will help performance for
a much longer time.

After you optimize the fdl, then
anal/rms/fdl=sysuaf.fdl sysuaf.dat sysuaf.dat

Make sure it stays in common and you don't put a separate copy in a root directory.

I'm always willing to follow up on the phone or email on any advise I give.

Have fun

Bob Comarow
VMS Consultant.
You can do a Purge/keep=2 to get out of the emergency.



If the security journal is huge, close it,
and start a new one.