Newsletter Q & A

The Q&A column is a frequent feature of the monthly newsletter sent to all MPE support customers. These questions are pulled from our support files and are real-life situations from our customers. They include both some of the more common questions and some of the more unusual problems handled by the Beechglen Support Staff.

1.) I am limiting who can log on to my system via telnet by utilizing the “allow” feature in the file INETDSEC.NET.SYS. I have a need to add more IP addresses. Is there a continuation character I can use to extend the ALLOW line to another?

The continuation character for INETDSEC is the backslash (). This is incorrectly documented as a forward slash in the HP Manual Configuring and Managing Internet Services. You can avoid continuations by making the file wider than the default of 80 characters and adding all of your address ranges on the same line. But, the best solution is to use better address ranges. Instead of listing each network individually like this:

telnet allow

Simplify it using this instead:

telnet allow 10.82.26-28.*

As you can see in the second example, you can use a hyphen to indicate a range of addresses, and an asterisk as a wildcard to denote “any” matching number.

2.) My VT users cannot connect to the system. In fact, they don’t get a logon prompt. Users connecting serially (DTC) and telnet users are working just fine.

You have reached the maximum number of VTSERVER processes allowed by the system. The default maximum is 300. Changed the maximum to a larger number with the NSCONTROL command..

:nscontrol status=servers


VTSERVER 0 300 300 0 OFF

:nscontrol start;server=vtserver,,500 (sets the MAX to 500)

Be aware that you may also have to increase the “Maximum number of connections” in NMMGR to accommodate the larger number of VT connections. This controls the maximum number of concurrent sockets the system will handle.

The NSCONTROL command to adjust the maximum number of VTSERVERs can be issued on the fly. You should also include it in your network startup procedures as it is not automatically retained. Changing the maximum number of connections in NMMGR requires restarting the network.

3.) I have heard about DBDRIVER during conversations and in articles I have read. What is DBDRIVER? How do I use it? There doesn’t appear to be any built-in help facility.

DBDRIVER is an Image intrinsic prototyping utility where you can call individual Image intrinsics in a command line fashion, without the need to compile a program. It is typically used to prototype Image calls during development, to test the behavior and results of a specific calling sequence for example. It is also used as a debugging and troubleshooting tool.

Beechglen maintains the TurboImage FAQ, located at . The FAQ has information on how to run DBDRIVER. Go to section 6.5 for DBDRIVER a commands reference. All throughout the text you will find examples of how to set up and call the different database intrinsics with DBDRIVER.

4.) I have several spool files that I cannot access, nor can I delete them. Whenever I try to access them the error message is


How can I delete these files?

To purge a file, a command or program must FOPEN a file and then FCLOSE it with a disposition of “purge.” It is the FOPEN that is failing with the FSERR 105. You need to use the PURGELINK command which doesn’t FOPEN/FCLOSE the file, thus avoiding the FSERR 105.


If that succeeds it will purge the file and unlink it from the spool file directory and you are done. If it fails, you will need to use FSCHECK instead.



The only problem with the FSCHECK method is that it will leave the spool file directory entry behind. The spool file will still show up in a LISTSPF or SHOWOUT command, but you cannot delete it. The problem will go away during the next system reboot when MPE rebuilds the spool file directory.

5.) Can a PC with Reflection be used as system console?

Absolutely. In fact, it offers several advantages over a dumb terminal. You can connect it to the network to utilize screen printing to a network printer, or you can e-mail it. You can turn on screen capture/logging to disk to record all console activity. Load software such as VNC or PC/Anywhere to allow you to remotely be the physical console from anywhere on your network. Imagine being able to reboot your system from home over a dial-up RAS or Internet VPN connection. Using a PC as a console is technically an unsupported configuration so it is a good idea to keep a real HP 700/9x terminal available to connect in the event there are console issues.

6.) What do the disk fragmentation charts in the monthly newsletter represent? My fragmentation is higher than 80%. Should I be concerned?

We are calculating fragmentation a little different than most others do. The chart is a ratio of the largest contiguous chunk of free space on a disk compared to the total free area. Similarly, it is also an illustration of how much of the total free area on the disk is *not* contiguous. For example, if a disk has 3 million free sectors, but the largest free area is only 500,000 sectors, we portray that as 83% fragmented because 83% of the free space on the disk is not part of the largest contiguous area.

I must admit that as disc drives get larger this methodology of calculating fragmentation becomes less meaningful. It is still useful for trending by offering a comparison from one month to the next to see if the chart changes drastically.

If you check the details with a :DISCFREE A command and you really are fragmented then you have several options.

  • Clean up disk space. Removing small numbers of unused files may seem to contribute to fragmented free space, but removing a large number of files could free up enough space to reduce fragmentation. If you would like we can perform a disc space analysis and look for disc space that you could regain.
  • Add more disc drives. Disc space is so inexpensive today that it can be more cost effective to add another disk than to purchase software to manage it.
  • Full Re-install
  • Run the CONTIGVOL command in VOLUTIL to condense all of the free space on a drive by drive basis.
  • Purchase one of the third party utilities to manage fragmentation. (De-Frag/X from Lund. Disk Space Manager from Bradmark).

7.) Are POSIX directories automatically on the system volume set?

HFS directories created directly off of root are indeed limited to the system volume set. HFS directories created underneath an MPE group will follow the home volume set definition for the group.

You can put a directory on a user volume by placing a symbolic link off of root to a directory that is underneath a group. For example, to create files in a directory named /datafiles, but you want to place them on the user volume PROD_SET use the following procedure.

:newacct POSIXFLS,mgr
:newacct POSIXFLS,mgr;onvs=PROD_SET
:newdir /POSIXFLS/DATA/datafiles
:newlink /datafiles, /POSIXFLS/DATA/datafiles

With this you can reference a file as /datafiles/filename and it will reside on PROD_SET.

8.) We no longer use Hewlett-Packard for hardware maintenance. Predict still runs (with errors) every night, but I have not been able to turn it off. How to I turn this service off?

You can disable PREDICT as follows:

:hello manager.sys
Select "4 - CONFIGURATION Options Menu"
Select "5 - SCHEDULE"
Select "1 - SCHEDULE (on/off)"
Enter "off" and press
Enter "exit"

9.) I have a question regarding sorted Image chains. According to the Turbo Image specification, if there is a tie on a sorted chain, the items following the sort item are used to “break the tie” and determine where in the chain the record should be placed. I have a database where some records were definitely out of order. Do you know why?

The items in the dataset immediately following the sort items, the ones that are used to determine placement on the chain when the sort items are the same, were DBUPDATEd after they were initially written to the dataset. Since neither the search item nor the sort item changed, there wasn’t a reason Image to resorted the chain. Only a DBDELETE followed by a DBPUT operation would guarantee sorting of the items in the application.

10.) We see this message on the console after some of our backups.


Can you tell me what it means?

The syntax of your store command is


Of particular interest is the “STORESET=” parameter. You are utilizing the “serial” STORESET feature of TurboStore. This means that the backup will write to LDEV 8 until it is full. Then it will continue the backup by writing to LDEV 9. This allows you to perform an unattended backup by mounting both tapes at the beginning with no operator intervention is required to switch tapes in the middle.

The console message is telling you that the backup completed entirely on LDEV 8 and it was not necessary to write any files to LDEV 9. Hence, the tape in LDEV 9 is not part of the current backup set.

11.) We have a new application that would benefit by having the Image database records sorted by index and perhaps DATE-ADDED. What are the implications to adding sort fields? The dataset tracks student registration and contains approximately 400,000 records.

The answer to your question about the implications of adding sort fields is ‘It depends’. When there are no sort fields on a detail set, records are physically added to the end of the file, or the next open slot on the delete chain, and logically added to the end of the chain for each path by updating the forward and backward pointers. Very little extra I/O is involved.

When a sort field exists, the physical location of the new record is the same, but the logical location (as represented by the pointers) must be determined by sequentially reading through the sorted chain until its correct location is found. Image will read backwards through the chain, making date-based sort keys a little more efficient, but a significant increase in I/O is possible.

Because of the additional I/O, there will be a performance hit. It is strongly recommended to only use sort items on short chains. As each chain grows longer more records may need to be read to find the correct location, resulting in more disk I/Os. I presume the key into the dataset is a student ID, and the number of adds and drops per student is not too high (<30), so I think you would not notice significant performance problems adding a sort field to your set.

One note about a DATE-ADDED sort key. The records by default (without any sort key) are chronologically added to each chain, so they should be in date-added order already, unless this field may reference earlier dates.

12.) How can I determine the MAC (Ethernet) address of my HP3000?

There are two methods you can use. If you are logged on to the HP 3000 with sufficient capabilities (NM capability required) you can use :LINKCONTROL @,CONF to display the “Current Station Address” for all configured network cards.

:linkcontrol @,conf
Linkname: DTSLINK Linktype: IEEE8023 Linkstate: CONNECTED
Physical Path: 56
Current Station Address: 08-00-09-22-96-2A
Default Station Address: 08-00-09-22-96-2A
Current Receive Filter: broad(1) any(0) k_pckts(1) x_pckts(0)

You can also obtain the MAC address from a Windows PC without logging on to the HP system as long as your are on the same logical subnet – no routers between your workstation and the system. Open a DOS prompt, ping the system (this method works for any system, not just HP 3000s), and view the ARP table.

C:>arp -a
Interface: on Interface 0x3000004
Internet Address Physical Address Type 00-aa-00-44-ff-47 dynamic 08-00-09-22-96-2a dynamic 08-00-09-ad-0c-f9 dynamic 00-50-ba-42-c9-95 dynamic 00-a0-c5-e1-88-09 dynamic

13.) When doing a partial store based on a date last modified, will all datasets of a database be stored even though they may not have been modified? Does store look to the root file access date for determination?

All of the datasets will be stored when you store the root file. TurboStore was enhanced to treat a database as a single entity. If you store the root file you get the whole database.

If you really only want to store the root file and the datasets that have been modified since a specific date, use the ;DATE= option and the ;PARTIALDB option of store. PARTIALDB tells store to treat each file individually, instead of the database as a whole, and compares the modify date in the file label with the date specified on the STORE command.

14.) I recently began using a system that had been turned off for a long time. The system appears to be running fine. But, my console repeatedly displays “WARN EF40 and WARN E040” in the lower left corner. What does it mean?

Older systems had internal batteries that would retain memory during power outages. This enabled these systems to survive a power failure without rebooting. Newer systems are connected to an Uninteruptible Power Source (UPS) to provide this functionality. Since the system was unplugged for such a long time the batteries discharged. The console messages you are seeing tell you that these batteries are recharging. Once the batteries are fully charged the status codes will discontinue.

15.) We installed additional memory in our system. Are there any tuning functions that might have needed to take place to utilize/optimize our new memory?

MPE automatically uses any memory that is installed in the system. No user intervention nor configuration is necessary.

16.) We are considering replacing our DDS-1 tape drive on our HP 937lx with a DDS-2 drive. Is there any software / sysgen configuration changes that will need to be made and if so what? Or is this just a hardware swap?

Just swapping the hardware will work just fine as both DDS-1 and DDS-2 drives use the same driver. But it is desirable to change the device id in SYSGEN to match the actual ID of the tape drive for support and documentation purposes. The SYSGEN ID for a DDS-2 drive is HPC1533A. Be advised that changing the ID will also set the DEVICE CLASS and MODE parameters to their respective defaults. Use the LIST DEVICE command to capture this information and apply it.

io> ld 7

io> md 7;id=hpc1533a;class=tape,dds;mode=autoreply

17.) Please let me know the command for doing a SHOWOUT on DEV=LP with any FNAME starting with P.

SHOWOUT really doesn’t have that level of sophistication. You really need to use the LISTSPF command as seen here.

:listspf @;seleq=[dev=lp and filedes=P@]

The SELEQ parameter of LISTSPF is very powerful. You can use any of the column heading seen in a “LISTSPF ;DETAIL ” command as selection criteria. Furthermore, wildcarding is acceptable.



The :SPOOLF command takes the exact same syntax so you can also select spool files to alter in the same manner.

:spoolf @;seleq=[jobname=backup and pri=1];pri=8;dev=laser2

18.) I am trying to load the OPENSSL package which I downloaded from HP’s JAZZ website. The instructions say to upload it as a ‘bytestream’ file. How?

FTP is by far the easiest method. Instead of using ASCII or BINARY mode, use TYPE TENEX as in the following example.

Connected to
220 HP ARPA FTP Server [A0009O15] (C) Hewlett-Packard Co. 2000 [PASV SUPPORT]
User ( doug.werth
230 User logged on
ftp> type tenex
200 Type set to L (byte size 8).
ftp> put filename ./filename
filename: File not found

Even easier is to place the downloaded file on a remote FTP server and “pull” it over to the HP3000. Use the “bytestream” command to set the transfer type.

File Transfer Protocol [A0009O11] (C) Hewlett-Packard Co. 2000 [PASSIVE SUPPORT]
220 FTP server (Version ready.
Connected to (FTPINFO 40)
Name(doug): doug
331 Password required for doug.
230 User doug logged in.
Remote system type is UNIX.
ftp> bytestream
200 Type set to L (byte size 8).
ftp> get filename ./filename

Lastly, if FTP is not an option, HP provides a utility to convert files into bytestream mode. Just upload the file using Reflection or MS92 in BINARY mode and then use TOBYTE.HPBIN.SYS to convert it to bytestream.

:tobyte.hpbin.sys "OPENSSL /tmp/openssl-vvv.rr.tar.Z"

19.) We are getting a message on the console “/etc/bootptab – NO SUCH FILE OR DIRECTORY.” What is causing this and how can I make it stop?

The console message is occurring because the HP has the BOOTP service enabled, but there is no configuration file. Each time it sees a BOOTP request on the network it goes to check its configuration file to determine whether or not to answer. You simply need to edit the file INETDCNF.NET.SYS and comment out the line for BOOTPS.

#bootps dgram udp wait MANAGER.SYS /SYS/NET/BOOTPD bootpd

Then tell INETD to re-read it’s configuration file. -c

You need to be logged on as MANAGER.SYS to modify the file and restart INETD..

20.) I am restoring all of my files from one system to another. How long should I expect the RESTORE to take?

The general rule of thumb is roughly 1 to 2 times as long as it took to create the STORE tape or even more. Many smaller files will take much longer than several larger files due to the overhead of allocating disk space, creating file labels, and updating the directory. Don’t forget to use the ;PROGRESS option of RESTORE.

21.) I’ve been told that if it takes around 30 seconds from the time you insert a tape cleaning cartridge until it ejects, that the cleaning process has been accomplished. Question is, is this true? Also when do you know if the cleaning tape is no longer cleaning?

That is more or less true. A cleaning cycle will last 30 seconds or so depending upon the model tape drive. A cleaning cartridge consists of a magnetic portion for data and a fibrous portion for the actual cleaning. There is a counter on the magnetic portion that is incremented each time the cartridge is used. When the maximum number of passes has been reached the drive will immediately eject the tape rather than run the cleaning cycle. This should be considerably faster than a normal cycle. Furthermore, if the “CLEAN ME” light on the drive is flashing before you begin insert a cleaning tape, it will remain flashing after the tape is ejected to indicate no cleaning took place. Many people will clean the drive on a regular basis even before the cleaning light comes on so this indicator doesn’t always apply.

22.) I rebooted my system last night. It stopped at the SHUT 4 message and would not completely shut down. I performed a hard reset (CTRL-B / RS). After coming back up the system was extremely slow for about 5 hours. Do you know why?

You are utilizing Mirror/iX to provide hardware redundancy for your data. Any time there is a system interruption such as a system abort or hardware failure the master and slave disks become out of sync. Mirror/iX has a known problem where the mirrors appear to be out of sync when a system does not reach SHUT 6. Consequently the system must rebuild the slave disks from the masters. This is an extremely CPU and I/O intensive process, especially for a large number of disks. Mirror/iX will rebuild no more than 6 pairs of disks simultaneously, but even a limited number of repairs have an extraordinary detrimental impact on system performance.

Don’t forget about our PRESHUT script. Run PRESHUT prior to system shutdown to give your system the best chance of reaching SHUT 6. If you don’t have the file PRESHUT.PUB.ESP you can download a copy from our web site at

23.) I need an Editor command file to search a file for a specific string and then delete the line immediately following it.

This sequence of Editor commands will accomplish the task.

/SET TIME=9999 */default is 50 iterations for 'while' loops/*
/ LIST *

24.) How can I tell if a remote session is connecting to our system via TELNET or VT-MGR. And if they are connecting through TELNET, what IP address they are using?

The SHOWCONN utility is most suitable for this task. Before using SHOWCONN for the first time there are a few commands you need to enter to set up the environment. You will also need to be a user with SM capability to use SHOWCONN.

:altgroup nsrv0000.telesup;cap=+PM

Once the above commands have been entered you can use SHOWCONN as follows.

:showconn job=#s35

#S35 MON JUL 1 2002 11:55 13 DEW,MANAGER.SYS,PUB 2723 vta utcind 88(VTSERVER.NET.SYS)

Total processes checked: 4
Processes with sockets : 1
Total sockets open : 1

25.) The normal ‘Vol (unlabeled) mounted on LDEV # 7’ message has not appeared for the past few days. Tapes go on-line and backups appear to work.

The system has multiple tape drives. A “:SHOWDEV 8” results in an a status of “UNAVAIL SYS”, indicating the other tape drive is “stuck” mounting a tape. The Automatic Volume Recognition (AVR) process, which handles tape and disk mount requests, is a single threaded process. That is, it can handle only one request at a time and all subsequent requests are queued. The AVR process is busy waiting on the tape mount request on LDEV 8 to complete. Inserting a scratch tape in LDEV 8 completed the request and all ensuing tape mount messages on LDEV 7 are once again appearing on the console.

26.) I created a handful of new fields on a form. “TABBING” functions do not work. I can use the arrow keys or mouse to place the cursor into any one of the new fields and enter data, but the TAB key bypasses all of the new fields.

All of the “bad” fields were incorrectly defined using a curly-brace “{” whereas all of the “good” fields start with a square bracket “[“. Change the form definition using only square brackets.

27.) My console is functioning fine. I can receive messages and type in commands. But I cannot get CTRL-A nor CTRL-B to work. Do you know why?

Check to make sure that LINE MODIFY mode is not enabled on the console. Press the USER/SYSTEM function key, then press MODES. Verify that the only mode that is enabled (indicated by an asterisk in the function key label) is “REMOTE MODE.

28.) If you replace a disk on a model A3550A do you have to do anything to repair it?

An A3550A is a Nike Model 20 disk array. If you are replacing a single disk that failed the disk array will repair automatically.

29.) Some of my users are unable to connect to HP system, what should I do? Local users work fine. Remote users are the problem.

Issue the command


to diagnose a gateway was flagged as disabled. Issue command


to fix the problem and explained the various possibilities why the HP networking software system ‘saw’ the router as unavailable and so marked the router as NOT active.

30.) How do you store to multiple tapes in parallel?

Here are the commands to use.

:file t7;dev=7
:file t8;dev=8
:file t9;dev=9
:store *t;;directory;show;progress;storeset=(*t7),(*t8),(*t9)

Be advised, you cannot RESTORE in parallel unless you own TurboStore Online or 24 x 7 backup. You must restore serially.

31.) I am receiving error messages on console that “mirrored partner is offline.” Is this a bad drive?

Replied to the console message to stop it from repeating. A DSTAT showed the drive missing. Placed a hardware call to have the disk drive replaced.

32.) A Job has been running too long. Also, we can’t abort a couple of users. Can you check into this?

Logged on and diagnosed a database lock causing both problems. Identified the offending session and aborted it. Once the lock was released the other sessions aborted and the jobs completed.

33.) My network group is requiring me to change the domain name of my system. What places need to be checked?

Check the following locations and change accordingly.

Machine Name
NS Node Name
DTC name(s)
NSDIR entries
Openview DTC Manager Workstation

Be advised that batch jobs, UDCs, and command scripts may have hardcoded domain names for systems that you connect to via FTP, DSLINE, etc. You can use the posix ‘grep’ utility to search all of your batch jobs for the old domain name and change them accordingly. Compiled programs could have the same problem.

34.) How do you set the tape request to automatically reply on an HP3000 running MPE 6.0? We need to automate the backup procedure. I think it is in SYSGEN but not certain.

The change is made in SYSGEN.

sysgen> IO
io> LD 7 (Verify this is the device you plan to change)
io> HOLD
io> EXIT
sysgen> KEEP
keeping to group CONFIG.SYS
Purge old configuration (yes/no)? YES
sysgen > EXIT

You will need to reboot for this configuration change to take effect.

For AUTOREPLY to function correctly the tape request must be unambiguous. For instance, if there are two devices with device class TAPE then you must set your file equation for a specific device number instead of device class. If you request a device class then you will still have an operator reply requesting which of the drives to use.

35.) Some running jobs are hung and others will not log on. Why?

A printer had stopped functioning. :SHOWDEV 137 reveals “UNAVAIL #J1437” indicating the printer is not spooled and has been taken “hot” by a batch job. As the printer is not working the I/O to the printer is hung, impeding the progress of the batch job. Issued an :ABORTIO 137 and :OPENQ 137 to allow spooling to the printer. Had to abort the job.

The system logon routine is a single-threaded process for all batch jobs. Once a job is “stuck” trying to log on all other jobs will back up behind it even if they are not waiting on any resources, limits, jobfence, etc. Once this job was aborted the other jobs were able to log on.

36.) I have some reports at the operators console that have been SPSAVED. For some reason I cannot find any information on printing them once they are in the SPSAVE mode. How do I do this? I need to change the device and test a new printer.

Use this command to produce a new spool file that is a duplicate of the SPSAVED file. The new spool file is given a new DFID number)

:spoolf O###.out.hpspool;print;dev=nn

37.) I am attempting to install two new disk drives. They failed to mount during boot up. I set them up in SYSGEN at SCSI addresses 8 and 9.

Ran ODE/Mapper to produce a hardware map of the system and found them at addresses 10/4/4.12.0 and 10/4/12.12.0. I recommended changing the addresses in SYSGEN to match the configuration of the drives rather than resetting the jumpers on the drives to match the SYSGEN configuration. Rebooted the system. Only one drive came up. The other had an LLIO error.

Check the pins/connector where the drive slides in. Since the drive does responds in mapper. When the controller responds but the disk does not it could be a bent pin on the SCSI connection or SCSI cable, or could just be a bad drive.

38.) I want a job number printed to $STDLIST at EOJ. How can I do that?

ECHO HPJOBNUM and HPJOBNAME and anything else you want to see. You could add these to a UDC named EOJ and every batch job would automatically execute them before logging off.

39.) My company’s security administrators want to disable SNMP on my HP3000. They believe that someone could use SNMP to disable the HP system. Is this true? If it is what can be done about it? How can we disable SNMP? Will disabling SNMP prevent any types of processing that we typically do?

First, the MPE version of the SNMP server has had performance implications and security issues. Make sure you have the latest SNMP patches installed. Second, the SNMP server is not enabled by default on a system. Use the :SNMPCONTROL STATUS command to see if SNMP is running. Before you can use any of the SNMP commands you need to enable the SNMP UDC files.

:hello manager.sys
:setcatalog SNMPUDC.NET.SYS;append

Disabling the SNMP server should not have any affect on normal processing. The SNMP server is strictly used by network monitoring devices such as Openview. While MPE network printing uses SNMP, it does not communicate with the MPE SNMP server. It strictly communicates with JetDirect devices using the SNMP protocol. Disabling the SNMP server on your system does not impact network printing.

40.) My backup job is hung. I am on the road and don’t have access to my system. Can you abort backup and restart network for me?

Logged on to the system and issued :ABORTIO 7 and ABORTJOB to kill the backup. Streamed his NETSTART job to restart the network services.

41.) In your February newsletter, you went over using the ZIP utility for the HP3000. I’ve tried to install but, not being a Posix guru, I’m having trouble just getting it on my system. I’ve downloaded from the ftp site, transferred from my PC to the HP3000 using binary mode transfer in Reflections. I then get into the Posix shell (and this is where it gets confusing) try to use the tar command (I’ve also tried downloading the mover version and using mover). I thought I’d just see if I can list the contents using the -t option of tar, but I get:

shell/iX> tar -x ZIPTARZ
tar: opening archive “/dev/tape”: No such file or directory

How do I get this loaded on my system?

I Emailed a WRQ Labeled format file of ZIP that you can just upload to an MPE/iX group. The group will need IA, BA, PH capabilities. If you would like some help going through the exercise of uploading the tar version I’ll be glad to help; but if you just want to get the program functioning then this will work.

The tar command to review the contents of the archive file should have been

shell/iX> tar -tvf ZIPTARZ

The -f option tells tar you are specifying a disk file named ZIPTARZ instead of accessing a tape drive.

Send mail to with questions or comments about this web site.
Copyright 2010 Beechglen Development Inc.