HiddenValue details commands and programs in MPE that can improve productivity on HP 3000 systems. Get a free NewsWire "HP 3000 Always Online" cap -- submit an MPE tip to Hidden Value. E-mail your tips to rseybold@zilker.net, or fax them to 512.331.3807.
Compiled by John Burke
Is it possible to add a disk drive "on the fly" without doing a reload?
Jeff Kell replies:
You generally have to shut the system down to install and cable the
disk to avoid
electrical/interface problems. The usual approach is to use SYSGEN to
configure the new
device on the path where it will reside, keep the new configuration,
shutdown the
system, install the disK and do a START NORECOVERY.
Once the disk is recognized by the system, you can add it to your running configuration as follows (assuming the new drive will be LDEV 5 in the system volume set):
:volutil
> newvol mpexl_system_volume_set:member5 5 90 90 (DISC,SPOOL)
[Editor's note: For details, see "Volume Management", HP Part No. 32650-90045 or "Performing System Management Tasks" HP Part No. 32650-90004.]
This will add the volume to the system volume set, but it also has some side effects. Since the new volume is "empty" and the disk space allocation routines attempt to "balance" loads across drives, all of your new files and transient space will be allocated on the new drive until it's capacity approaches that of the other volumes. This will create an I/O bottleneck on that drive, at least initially.
You could selectively :RESTORE certain accounts (or the whole system) to try and balance the allocation. You could also perform an INSTALL and a :RESTORE for better efficiency, but at the cost of a great deal of time. There are also certain third-party utilities that will balance disk utilization across members of a volume set. These utilities work online on a running system and don't require any downtime. [Editor's note: See this month's Test Drive for a discussion of these utilities.]
When I do a FORM SETS command from QUERY on one of my databases, it shows a dataset with a negative number of entries. How can this be?
Evan Rudderow replies:
It's been my experience that when you see negative numbers in
something like a
FORM SETS it's because someone, in an attempt to recover from some sort of
failure,
restored only some of the datasets rather than the entire database and
there had been a
dataset capacity change.
Adager's Ken Paul elaborates:
Many people are surprised to find out that IMAGE does not keep the
entry count
for each dataset anywhere within the database. The entry count for a given
dataset is
actually the result of subtracting the FREE entry count (kept in the
dataset's userlabel)
from the dataset's capacity which is in the root file.
If you have a dataset with a capacity of 5000 and 3000 entries within the dataset, the free entry count is 2000. Then someone changes the capacity to 10,000 and the free entry count is changed to 7000. If someone else were to restore the OLD root file, then IMAGE/DBINFO/QUERY would report the number of entries as -2000 because 5000 - 7000 = -2000. [Editor's note: Adager and DBGeneral can both detect and correct these problems.]
The network configuration of our HP 3000 was originally set up with one block of IP addresses. Now I need to add another block of addresses. Where do I add these in NMMGR?
T.N. Anand replies:
You can add an IP address using NMMGR the following way:
€ After typing NMMGR, select "Open Directory" .
€ Then select "Update Dir."
€ Now select the "Add" option (F5.)
€ You are placed in a screen where you can enter the IP Address of the
machine. The
type is generally set to 1( IP).
€ Now press the "Save Data" (F6) option, back out of NMMGR, and you
are done.
How can I obtain the current time zone value programmatically -- you know, the value shown in the SHOWCLOCK command output?
Jeff Vance replies:
I named my command file "tz": [Editor's note: normally we do not
include code
snippets, but this one is short, and, considering the author, studying the
technique is
probably even more valuable than the command file itself. Also note that
rht with a
negative count only works on MPE/iX 5.5.]
parm entry=main
# script that parses out the time zone from SHOWCLOCK output
if '!entry' = 'main' then
errclear
continue
showclock >./shclktmp
if hpcierr = 0 then
xeq !hpfile entry=doit <./shclktmp
deletevar _tz_@
endif
return
else
# 'doit' entry:
# $stdin comes from showclock command redirected from ./shclktmp file
# discard 1st three records. The actual TZ display consists of 2
# records: the number of hours, followed by the description
input _tz_rec
input _tz_rec
input _tz_rec
input _tz_rec
setvar _tz_rec ltrim(rht(setvar(_tz_rec,_tz_rec+input()), &
-(pos(":",_tz_rec)+1)))
# now _tz_rec contains the TZ portion :showclock output
echo !_tz_rec
return
endif
How do I eliminate Posix files from a backup?
Stan Sieler replies:
For MPE/iX STORE, you can for example do: STORE ?@.@.@
We recently replaced 18 Eagles and Coyotes with 6 FW SCSI drives. We freed up 12 LDEVs in the process. However, some LDEVs still list with the SHOWDEV command and some LDEVs don't list just as I would expect. What's up with that? Also, what is the recommended number of Gb per FW SCSI interface card?
Gilles Schipper replies:
Presumably, you deleted the "freed up" LDEVs with SYSGEN. However,
even if you
did the right thing by deleting the unused LDEVs via SYSGEN, these LDEV
would (and
should) still appear on the showdev output. In such a case, they would
simply represent
"pseudo" logical devices potentially utilized by NSVT and/or telnet.
As to the recommended number of GB per FW SCSI interface -- I'm not sure that is as important as the number of spindles per interface. Although the theoretical maximum is 15, I'm not sure I would want to have that many on a single interface. [Editor's note: based upon previous threads, the consensus is that a maximum of 10 devices should be placed upon one FW SCSI interface.]
We are setting up a firewall to proxy NS/VT service to the HP 3000. By using port 1570 NS/VT works fine, but we need to know the TCP service port HP uses for network printing under MPE/iX 5.5 to allow printing to pass through the firewall. Any ideas? We are using a JetDirect box with an IP address assigned.
Larry "MPE/iX Spoolers 'R' Us" Byler replies:
The TCP port number to the JetDirect interface is a configurable option in
NPCONFIG (useful for JetDirect EX+3 interfaces, which have three such
ports). The
default, if you do not specify otherwise, is 9100.
Is there a limit to the number of digits a job number can contain (HPJOBNUM) on MPE/iX?
The HELP text for :SETCOUNTER reveals:
"...MAX must be less than or equal to the maximum possible value for
that counter.
Those values are: JOBNUM --16383; SESSNUM --16383" So it appears
"yes, five" is the
answer.
After adding an additional network to a neighborhood gateway in NMMGR, is it necessary to reboot the system for the network to be recognized? Or is there another way to make the system recognize the changes that were made?
Doug Werth replies:
You can update "on the fly" with a
:NETCONTROL UPDATE=INTERNET;NET=[name of your network, typically LAN1]without rebooting the network or the system.