RAC Sponsor Message

Hidden Value details commands and procedures in MPE that can improve your productivity with HP e3000 systems. Send your tips or comments to john@burke-consulting.com, or fax them to 512.331.3807.

Edited by John Burke

I need to add storage to my HP 3000. It appears that many people put standard Seagate or IBM or other SCSI drives on their HP 3000 with no problems. I’ve got both SE and FW interfaces on my HP3000 987. Is the FW interface LVD or HVD? I have several 36Gb LVD drives but I don’t know if they’ll work on my HP 3000.

Denys Beauchemin replies:
The SE connection is a FAST SCSI-2 Single Ended connection. The “FW” connection to which you refer is actually a Fast-Wide Differential SCSI-2 connection. You can only add HVD disk drives to the latter. I have used a fairly extensive variety of SCSI disk drives in the systems here, and I like both Seagate and IBM drives. The difficulty will be finding HVD disk drives. These seem to be out of production everywhere as the industry has gone over to LVD in the UltraSCSI, UltraSCSI2, Ultra160 and Ultra320 guise. However, you should be able to find these drives on the open market. I believe the largest HVD drive is 36Gb, but I may be wrong.

You can also buy the LVD drives and connect them to your SE SCSI port, but that would be a waste of high performance hardware, as the interface will only work at the lowest speed on the chain. Another idea would be to get an HVD to LVD converter, and such can be found at www.paralan.com and other sites.

How do I check whether a particular listener job is currently accessing a database?

Jon Diercks replies:

LISTF database,8>$NEWPASS
/bin/grep ‘-q JOBNAME’ <$OLDPASS
IF CJCW = 0 THEN
BINGO
ELSE
NOTBINGO
ENDIF

I copied the configuration file but cannot use the command chown to change owner. I’m logged on as MANAGER.SYS. What am I doing wrong? Here’s what I’ve tried:

shell/iX> cp /SENDMAIL/PUB/sendmail-8.9.1-m/cf/cf/mpeix.cf /etc/sendmail.cf
shell/iX> chown SERVER.SENDMAIL /etc/sendmail.cf
chown: file “/etc/sendmail.cf”: Invalid argument

Guy Paul and Mark Bixby reply:
The existing file probably doesn’t have a group owner assigned., If that is the case, for some reason MPE won’t allow you to change the owner unless you also specify the group owner as well. Do this command:
chown SERVER.SENDMAIL:SENDMAIL /etc/sendmail.cf

I inadvertently typed the name of my HP e3000 in the following command on my desktop PC: C:\WINDOWS\Desktop> net time \\HP979. It returned the following display:
Current time at \\HP979 is 5/9/2001 hh:mm
The command was completed successfully.
I saw that hh:mm is the same as that displayed by SHOWTIME on the 979. Adding the parameters /SET and /YES to the above command makes my PC clock = the 979’s clock. Is this time being returned by Samba, or is it being supplied from somewhere else?
.
John Clogg replies:
Yes, Samba is responding to the NET TIME request. [Editor’s note: See this month's net.digest for more neat Samba tricks.]

I have a client with a brand new N440. On one drive the device ID is reported differently between DSTAT and SYSINFO. DSTAT calls it LDEV 40, and says it’s MAJ3182MC. SYSINFO calls it ST318404LC. Does it really matter?

Stan Sieler replies:
Your MAJ3182MC is a Fujitsu 18.2Gb Ultra3 SCA-80pin SCSI HDD, a 10,000 RPM, 4.7-millisecond average seek time disk drive. The ST318404LC is a Seagate 10,000 RPM, 5-millisecond access time disk drive.

SYSINFO always reports the device ID you configure, which may or may not be the correct device ID. DSTAT ALL tries to report the actual device ID.

As it happens, you can tell SYSGEN that ldev 40 is an “MAJ3182MC”, at least on 7.0 Express 1 (i.e., that device ID is in IODFAULT.PUB.SYS).

I recommend having SYSGEN’s configuration reflect reality, where possible.

I zipped up a few files using PKZIP for Windows on an NT 4.0 machine, FTP’d the archive file to one of my HP e3000s and tried to unzip it with
SYS-C:run /SYS/HPBIN/UNZIP;info=”-l /SYS/HPBIN/z0test”
but I get a long error message that basically says it cannot find the zip file directory. What am I missing?

Neil Harvey replies:
I have seen this often, and it’s usually because I didn’t FTP the file across properly. When FTP-ing a zip file from NT to MPE, I always try to remember,

hash
bin
quote type L 8
put c:\nt.zip /tmp/nt.zip
quit

I have a command file where the user enters in the dfid number of a spoolfile. If it’s there, it displays correctly. If it is not there, the command file aborts. How do I trap the fact that the file does not exist and start over? Here’s my attempt:

ECHO PLEASE INPUT THE DFID NUMBER
INPUT DFIDNUM
LISTF O!DFIDNUM.OUT.HPSPOOL,0
IF CIERR <> 0
ECHO NON-EXISTENT FILE - TRY AGAIN
ELSE
...display file

Robert Mills replies:
Use the finfo function as in
ECHO PLEASE INPUT THE DFID NUMBER
INPUT DFIDNUM
if not finfo(‘O!DFIDNUM.OUT.HPSPOOL’,’exists’) then
ECHO NON-EXISTENT FILE - TRY AGAIN
ELSE
...display file

There are five date/time stamps stored on each file label: created, modified, accessed, allocated and status change. The first three can be seen with :LISTF,3 and the other two can be accessed via FINFO(). What are the specific rules for when the alloc and/or status dates are updated? In general trial and error, “allocated” usually matches “created” and “status change” usually matches “modified,” but for both I have seen cases where they differ.

Lars Appel and others add:

The “allocated” timestamp reflects when the disk space was allocated for the file. This is usually the same as the “created” timestamp, but one example of where they can be different is in the case of :RESTORE;OLDDATE — the create date will show the original create date, and the allocated date will show the restore date.

The “status change” timestamp is required by Posix. From the Posix perspective, file status is modified when the file’s Owner or GroupID attributes are altered, when file permissions or link count are changed and (for directories) when an entry in the directory is added or removed. To support these requirements, MPE added a new State-Change timestamp for all files. One example command that causes “status change” to be different from “modified date” is :/bin/touch ‘-a <filename>’