Q: My system seems to allocate ldevs 90-99 to something called VTERM and I would like to use some of these numbers for network printers. Does anybody know how to free up these ldevs or re-assign VTERM to different numbers?

A: Doug Werth replied:

As far as I know you cannot free them up while the system is running. You can, however, configure your network printers in SYSGEN to use the 90 thru 99 range and they will work on the next reboot. The SYSGEN devices are allocated earlier in the bootup sequence than the VTERM devices.

Q: I am terribly short of disc space on LDEV 1 as the following DISCFREE listing shows:

          | Configured         | In Use             | Available          |
----------+-------------------+-------------------+-------------------+

LDEV : 1 -- (MPEXL_SYSTEM_VOLUME_SET:MEMBER1)

I believe I can use VOLUTIL to change the maximum from 75% to 95% and that will allow my available permanent space to increase. Is this a good thing to do? If it is a good thing then should I set it to 95% or can anyone suggest a better value? Should I do a system store first? Can it be done without a reboot?

A: Lee Gunter (in a two year old posting), Jon Diercks, Gregory Stigers and John Burke contributed to the answer:

Can you increase the maximum to 95% using VOLUTIL? Absolutely. Should you do a system store first? Not necessary. Can it be done without a reboot? Absolutely.

Now for the hard part, is there a better value? It depends primarily on how you use the system. If you were to configure permanent at 95% and then proceed to use it all up, you would only have 5% for transient space (note you were using 6% at the time of the DISCFREE). Since you only have this one disk in the system volume set, limit yourself to at most 85% allocated to permanent. Better yet, remove all unnecessary files until you can address the issue by adding one or more disks.

Q: I cannot run my ODBCJOB.ODBCSE.SYS job because it cannot purge the log file. How can I purge this file? I logged on as MANAGER.SYS and even tried to rename the log file but encountered the same error:

:PURGE ODBCLOG.ODBCSE.SYS
BAD VARIABLE BLOCK STRUCTURE (FSERR 105)
Unable to purge file "ODBCLOG.ODBCSE.SYS". (CIERR 384)

A: Doug Werth replied:

The PURGE command FOPENs then FCLOSEs the file with a disposition of "delete." The FOPEN is what is failing due to the file system error indicated. To purge the file you must utilize a method that does use the FOPEN/FCLOSE combination. Try one of these options instead.

:purgelink ODBCLOG.ODBCSE.SYS

:rm.hpbin.sys /SYS/ODBCSE/ODBCLOG

:fscheck.mpexl.telesup
>purgefile ODBCLOG.ODBCSE.SYS
>exit

[Editor's note: the last should be undertaken with great care and only if the first two suggestions fail.]

Q: I am looking for a suggestion on how to trap a DBUTIL error in a batch job. I find nothing in the manual about this. I've tried making errors and checking JCWs and variables, but nothing turns up.

My particular need is to restore part of a data base to another system and DISABLE database FOR INDEXING. It would be big help if the $STDLIST did not need to be checked manually to verify that this step was successful. Any ideas?

A: Michael Berkowitz replied:

The CIERROR JCW will be non-zero if the DBUTIL command fails.

Q: Has anybody had success trying to split a single HP3000 system into two different OS versions by using two sets of discs? There is no need for sharing of information between the two OS'es and they would run independently of each other at different times of the day.

A: Guy Paul replied:

This is certainly possible, as we have done this in the past for customers who couldn't tolerate any downtime for OS upgrades. Hence we came up with a solution to have a duplicate set of SYSVS discs that we upgraded while they were still on the old OS. Come day of the 'real' OS upgrade, brought them down, stored off any modified files, switched over to the new OS, restored modified files and they had an OS upgrade in about 45 minutes. So it is possible.

You should probably consider using BULDACCT to synchronize the accounting structure.

Gilles Schipper added:

This should be entirely possible. I do this sort of thing all the time. By simply booting from the appropriate boot path, you can do exactly as you wish. In fact, I have even shared common volume sets among different LDEV 1 system volume sets, with different MPE versions.

Q: We just installed a 100Base-Tx card in our 939ks. Now a job used to extract data from an Image database into SQL Server 7 has gone from an average of 1 hour to over 6 hours to complete. Also, when using Reflection to download, the progress bar will show about 10% of the file downloaded, then pause, then another 10%, then pause, etc. The card is set at full duplex with autosense off and is connected to a switch. Does anyone know what might be going on?

A: Greg Skvorak replied:

If you haven't already, try forcing the switch port to 100mb. It can't hurt, I've seen it fix similar problems, and, best of all, it doesn't take a reboot of the network on the 3000.

Q: If I perform a LISTF fileset,2 to a disk file in a session, the first file appears in record #6 while in a batch job it is record #7. The difference is that the batch version has a date and time stamp in the first record. This bit me when I transferred a working command file that had been executed only on-line to a batch job. Is this a bug?

A: John Burke replied:

Actually, this is another reason you should forget you ever knew the command LISTF. Use LISTFILE instead - it behaves the same way interactively and in batch. LISTF will probably never go away, but it also is unlikely to be "fixed" since backward compatibility has always been important and LISTF has always worked the way you describe. Remember that LISTF dates from long before the time you could write CI scripts like today.

Q: I've been wondering to myself lately "What are the consequences of not taking down the network during our full backup?" Having the network up is a big help with some of the automation projects I'm working on. What about when creating an SLT?

A: Gilles Schipper replied:

Leaving the network up should have no undesirable consequences on a full backup. You do not need to worry about any critical system files being omitted from the backup. However, by leaving the network up, you could be exposing the system to users logging on and in turn opening files, thus risking backup integrity. That can be avoided by ensuring the jobfence is raised to prevent unauthorized logons. This is generally preferable to bringing the network down.

As for creating an SLT, I have done an INSTALL many times from an SLT tape created while the network was up - absent any negative consequences. In fact, I encourage my customers to create SLT tapes regularly. And nothing discourages creation of an SLT more than the requirement that it be completed while the network is down.

Q: I have looked through the 6.0, 6.0pp1, 6.5 and 6.5pp2 communicators and can find nothing about the checkpoint improvement code, as mentioned in the help command for VOLUTIL. My questions What does it improve? Any reasons not to turn it on? Does it really make any difference?

A: Bill Cadier replied:

The check point improvement code operates by keeping a bit map of the portions of a file that changed so only those portions need to be posted by transaction management. This option doesn't exist in 6.5. The bit map method does not scale well with > 4GB files so it was removed.

Q: I seem to remember that there is a command within FTP for MPE/iX that allows you to stream a job. But I do not remember the syntax. Can anyone help?

A: Ronald Horner, Jonathan Backus, Andreas Schmidt and Gibson Nichols replied:

SITE STREAM filename

Q: What is the command or program that displays the amount of memory on a 3000 system?

A: Fred Metcalf replied:

SYSINFO.PRVXL.TELESUP

Q: I just purchased a 928LX as a development box. Has anyone seen this error at Boot Up? It's as if I have no boot device.

Main Menu: Enter command or menu > BO
Interact with IPL (Y or N)?> Y

Booting...

ENTRY_TEST status = 0
Failed to initialize
ENTRY_INIT status = -4
0B300040 00000000 00000000 00000000 00000000 00000000 00000000 00000000
00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
00000340 0000BF40 00000000 00000000 00000000 00000000 00000000 00000000
00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000

A: Russ Smith replied:

This usually means you don't have a SCSI terminator on the multi-I/O card. If you do, try reseating the ribbon cable leading to your drive cage.

Q: When you change the IP address for a device already configured in NPconfig, do you have to take the network down and restart it to force a read of the new IP address?

A: Christian Lheureux and Fred Metcalf replied:

No. Just stop and start that ldev's spooler.

Q: How can I print a spoolfile that is defered with a status of SPSAVE? I have copied the file, then tried to print, but the control characters aren't coming across.

A: Richard Bokal, Gordon Montgomery and H. Lassiter replied:

SPOOLF Onnnn.OUT.HPSPOOL;PRINT;DEV=ptrname;PRI=n;COPIES=n

Q: I have been asked to determine the configuration of the NIC card on my system. Can anyone point me in the right direction?

A: Richard Bayly provided the easiest, and also best, answer:

nettool.net.sys "conf;summ;quit"

Q: Is there some MPE command or system utility that shows who is logged on over an FTP connection? Something like a :SHOWJOB listing?

A: Richard Bayly and Lars Appel replied:

Use SHOWCON or, if on MPE/iX 7.0, use LISTFILE FTPSRVR.ARPA.SYS,9 to get a list of the server process PIN numbers and the "fmtuid" or "fmteuid" option of PINFO() to see the PIN's "user.acct" info. You can also use PINFO to find the IP address, as in PINFO(pin,"IPaddr").

Q: Is there any command to show the file links that have been set on a system? Something along the lines of LISTEQ, but for files, both link and target, which have been set-up with a NEWLINK command.

A: Mark Bixby replied:

If, for example, you want all the links and targets under /etc, then from the CI,

listfile /etc/,unique;seleq=[object=symlink]

And, from the shell,

find /etc -type l | xargs ls -dl

Q: We ran out of disk space over the weekend. Now my JOBQ's are screwed up. When I do LISTJOBQ for HPSYSJQ, it shows 6 EXEC but only a limit of 3 and a total of 3. When I do a SHOWJOB, only 3 jobs show for this queue. I'm having the same problem with other queues also.

Any way to fix this without bouncing the system?

A: James Tollerson, Larry Simonsen and Richard Bayly replied:

Basically no, though you can adjust limits to achieve the desired results until such time as you can reboot. There is a patch for this problem:

MPELXC2B - 6.5
MPELXC2D - 7.0
MPELXC2C - 6.0 but superseded by MPELXL7A