ROC Software Sponsor Message

Hidden Value details commands and procedures in MPE (and some in Vesoft’s MPEX) that can improve your productivity with HP 3000 systems. Get a free NewsWire HP 3000 Always Online cap — submit your MPE tip directly to us here at the NewsWire. Send your tips to editor@3000newswire.com, or fax them to 512.331.3807.

Edited by John Burke

We just purchased a used DDS3 to replace our DDS2 because we were told it would reduce backup time. Well, the backup did work, but it didn’t take any less time. I only have 120-meter DDS2 tapes. Is this the problem? Will they work in this drive? Also, how do I configure the drive for compression?

Denys Beauchemin, Stephen Douglas and Neil Harvey reply:

If you want to write in DDS-3 format on a DDS-3 drive, you MUST use 125-meter tapes. If you want to write in DDS-2 format on a DDS-2 drive, you MUST use 120-meter tapes. The DDS-3 drives are able to read DDS-2 and DDS format. The reverse is not true.
With a DDS-2 drive and 120-meter tape, you can get 512 Kb/s to 1 Mb/s in transfer rate. You get the same with a DDS-3 drive and 120-meter tapes. If you use 125-meter tapes in your DDS-3, my documentation shows a transfer rate between 1 Mb/s and 2Mb/s, depending on the compression.
Also the capacity on DDS-2 tapes is 4-8Gb (DDS-2) depending on compression, versus 12/24Gb on 125-meter tapes (DDS-3). If you use 90 or 60-meter tapes, you are using DDS format, which on a 90-meter tape will store 2-4Gb.
You can turn the DDS drive’s compression on (or off) by doing the following:
DEVCTRL.MPEXL.TELESUP n COMPRESSION=ENABLE (or DISABLE)
where n is the LDEV number of the DDS drive.

Is there a way to enable or disable system-logging events without rebooting the system?

Allegro has a free utility available at their Web site (www.allegro.com) called ‘SYSLOG’ which allows you to do this. You’ll find it in their ‘free software’ section along with a myriad of other useful programs.

Apparently, I have an invalid argument in INETD.CONF. I would like to view the file and compare it to another HP 3000 system. Unfortunately I can’t locate the file. I suspect I have to go into Posix. What do I do?

Denys Beauchemin, Gerald Dillard, Mark Klein and Andreas Schmidt reply:

From MPE (assuming the links are set up correctly),
print inetdcnf.net.sys
If /etc/inetd.conf is not a link to INETDCNF.NET.SYS then you’ll need to look at it directly. Again, from MPE,
print /etc/inetd.conf

I need to restore a few files from some backup tapes that I have, but need to know the procedure for checking when the data was stored on the tape. I’ve tried: RESTORE *T;LISTDIR. This gives a listing of directories, but not the store date. Is there another commend that I need to add to this?

Kevin Miller and H Lassiter reply:

If you provide a wildcard, you get the media info.
:RESTORE *T;@;LISTDIR

Where can I find some documentation on configuring/testing the internal modem? I called the Response Center and was told to use modmutil and was even e-mailed instructions for it, but it requires a support license password (why is a modem utility so dangerous that it needs a password)? Is there another way? What I want to do is create a pager job that can use the modem to dial a support pager if something goes wrong in our nightly production.

Rick Clark replies:

I have created the same type of job on my system. However, after testing I found that a modem connected to a DTC port would work the best. There are two reasons that I can think of off-hand. Reason one: never use the support modem other than what it is intended for; and two: you will have to use the command REFUSE on the port. This disables (I believe) any inbound sessions.

Doug Werth added:

I absolutely agree with Rick, it is much better to connect a modem to a DTC. Furthermore, a modem connected to a DTC can be power-cycled to reset it, as well as using TERMDSM to reset the port. TERMDSM does not require a password from HP, while CONSOLAN and MODMUTIL do.
Another place that I prefer external devices is DDS drives (whenever feasible) for the same reasons. Shutting a system down to reset an internal DDS drive that won’t eject a tape stinks. Often, simply cycling power will release the tape.

I’m trying to use the new job queue feature of 6.0. Can the limit command be used to limit the number of concurrent jobs that will run in the system queue as it does for other queues? What I want to do is to set a global limit of 4 with the system q limited to 1 job, the daemon queue limited to 2 and another job queue (reserved for me) limited to 1. Also, how do I get the queues to come back after a START NORECOVERY?

Jeff Vance replies:

Try LIMIT nnn;jobq=HPSYSJQ. The whole process would be:
LIMIT 4
LIMIT 1;jobq=hpsysjq
LIMIT 2;jobq=daemonq
LIMIT 1;jobq=myq
This can be part of a stream job that can be launched from sysstart [as noted by Barry Lake].

I am trying to write a program that reads records in a KSAMXL file based on an exact match to a key value. The file can contain multiple records based on the key value. I only want to read those records that match the key value exactly. I have been able to successfully find and read the first record using FREADBYKEY and FFINDBYKEY+FREAD, but then subsequent FREADs will take the rest of the KSAMXL file until it hits EOF. How do I limit the FREAD so that it only reads the qualifying logical records (based on the key value)?

Costas Anastassiades and Tony Summers reply:

You can’t limit it through the use of intrinsics. Use FFINDBYKEY to point to the first record (using a relational operator of equal). After each FREAD compare the value you just read to the key value you specified for FFINDBYKEY (and stop FREADing when it changes).
Reading a KSAM while other users are writing or updating can prove tricky. Obviously FLOCKing ensures you are reading what is there because no-one can write or update the file while you have it locked. However, FLOCK actually does two things I believe. It locks the file AND it flushes the buffer and it’s the flush which actually ensures you get any records added by other users as it forces the next read to perform a physical IO. So instead of FLOCKing we regularly flush the buffers (FCONTROL 7).

Is there a limitation on logon UDCs? On our system we have two system level UDC files, one for VEsoft’s products and one of our own. I tried to add a logon UDC to our UDC file, but it does not get executed at logon, while the VEsoft logon UDC does. What am I doing wrong?

Stan Sieler replies:

Yes, there is a limit. At each level (system, account, and user), only the “first” seen logon UDC will be invoked at logon time.
Indeed, you could well argue that SETCATALOG ought to warn you if there is one or more UDCs marked as “option logon” that won’t get executed due to this restriction. Alternatively, perhaps we should request a functionality change, to allow multiple logon UDCs at each level.

Can you have a series of 4Gb drives and 9Gb drives that are mirrored in the same volume? So long as the source drive and its mirror are the same size?

Kelly Tomlinson and Claire Pope reply:

Yes, definitely.

Steve Cole adds:

Issues with mixing disk drives of different sizes can occur at different times. The issue is that if you add 9Gb drives to a volume set with 2Gb or 4Gb disks (with no reload) and the environment is reasonably static, then the data that you access is still spread across all disk drives reasonably well. However, if you reload the volume set you must keep in mind that MPE/iX writes data to the disk drive on the volume set (excluding LDEV 1, which is a special case) with the most available free space. This means the first data restored will not be spread but will be written to the 9Gb disk(s). After the free space on the 9Gb disk(s) has been adequately utilized, the remaining data is spread across all of the other drives. This can create a potential I/O bottleneck with the 9Gb disk and impact performance.
We observed this on a system that was upgraded to 6.0 by means of an install. After the upgrade was completed and all the data was reloaded, the system ran significantly slower. We were called to evaluate the performance issue. When we dialed in we found that 90 percent of the user data was concentrated on the one 9Gb drive and the system was suffering from a severe I/O bottleneck.
Depending on the I/O demand on your system, mixing disk drives of different sizes may cause performance issues.

We want to use a printer connected to Intel’s NetXport print server boxes as a network printer from the HP 3000. Can we do this? We have done this with printers using the JetDirect card, but Intel’s boxes have two printer ports. There doesn’t seem to be a way to specify what printer port on the NetXport should be used by the HP — and apparently there is no default.

John Burke replies:

It’s been a little while since I tried this, but it did work. In the npconfig.pub.sys definition for the printer, add the following two lines:
snmp_enabled=false
tcp_port_number=n
where n=2501 for the serial port and n=3001 or 3002 for the parallel ports.

Somewhat randomly, we get a handful of heartbeat losses, carrier losses and transmit errors (same number of each). We can easily go for days without seeing any more. We replaced the MIO card some months ago for a different reason, but it seemed to have had no effect on these occasional glitches. I’d like to replace the transceiver because we see no other problems anywhere on our network. What are my chances of successfully doing this hot?

Cory Black, John Lee, John Skelton, Tim Manns and Jean-Paul Schmit all report successfully swapping transceivers hot. In fact, Tim Manns reported that replacing the transceiver solved the problem exact problem noted in the question on one of his systems.

The following is the only way I can find to set our tape drive (LDEV 8) to autoreply.
:SYSGEN
SYSGEN>IO
IO>MDEV 8 MODE=AUTOREPLY
IO>HOLD
IO>EXIT
SYSGEN>KEEP
SYSGEN>EXIT
I create a SLT then reboot. Is this correct? And on a full system backup should I use the following commands?
:FILE T;DEV=8
:STORE @.@.@;*T;SHOW=OFFLINE
I thought a ‘/@’ or something was also required to get all of the Posix files?

Stan Sieler replies:

Yes, but you don’t have to boot (“UPDATE”) from the tape. Simply reboot as “START NORECOVERY” (assuming your config group is “CONFIG”) will load the revised I/O configuration. However, note that it’s always a good idea to generate the SLT tape anyway.
As of 5.5 (maybe earlier?), STORE @.@.@ is treated like “STORE /”.
However, we do:
store / - /SYS/PUB/NL - /SYS/PUB/SL - /SYS/MPEXL/DUMPAREA &
- /TELESUP/DUMPS/@ - /tmp/ ; ; &
compress; onerr=quit; partialdb; directory; show=long,dates; progress
The “- <filename>” stuff causes the backup to be shortened by many Mbs of data ... data that either comes in from the SLT tape (NL, SL, DUMPAREA) or that we don’t care about (memory dumps, temp files in /tmp/).