Q: I ran checkslt on the MPE/iX 7.5 SLT and
it failed. It failed on a DDS2 drive on two different systems but passed when a
DDS3 drive was used. The MPE/iX 7.5 SLT is on a 120 meter DDS2 tape. HP, at
present, is looking into this. Has anyone experienced this?
A: Michael Berkowitz replied:
What makes you think you don't have two bad
DDS-2 drives. When we had them, we went through them like water, replacing them
every couple of months. They are bad news from the word go.
After the original poster balked at the idea
he could have two bad DDS-2 drives, Gilles Schipper noted:
Not surprising at all. I once experienced
the following situation. Our customer had a disk crash. Fortunately, it
happened just after a full backup. HP replaced the faulty disk drive and we
proceeded to perform a system reload from the just-completed backup that had
been to a DDS-2 tape drive.
As soon as we mounted the tape (on exactly
the same tape drive that created it), we received a console message indicating
AVR error on ldev 7. I knew right away we had a problem. HP returned to replace
the tape drive with another DDS-2 drive. Still no joy. We recommended replacing
the drive with a DDS-3 tape drive. As soon as this was done, the reload
proceeded without further problems.
The bottom line is, as has been mentioned in
another post, stay away from DDS-2 drives, as far away as possible. From this
experience and others, I have concluded that the dds2 drive is, to put it
mildly, flaky.
Q: This morning I came in to find the backup
job stalled. Abortjob was ineffecive, as was abortio. I ended up rebooting the
system. While coming up, I got the "defective sector" message with
"FILE.GROUP.ACCOUNT has an extent with unreadable data". The file is
now locked and I need to use FSCHECK to unlock it.
How can I determine which drive this extent
is on? I have a good idea which one it
is, but I'd like to be 100% sure before I replace and reload.
A: Stan Sieler replied:
FSCHECK's "DISPLAYEXTENTS" command
may help. Note that, if I recall correctly, it displays logical unit numbers,
not exactly LDEVs.
Q: We were going over our time change
procedures and came up with questions about all the different time change
methods and parameters and what they mean. For years upon years we have
automatically run a job on the time change Sunday at 1:59 AM that basically
just changes the time zone. For the last couple of years we have also started
running an NTP time synchronization job at 3:00 AM every morning. The TZ
variable is set to "PST8PDT". Now as I understand it, NTP is
adjusting the UTC time. But does it also adjust the time zone? In other words
are we duplicating effort with these jobs?
A: John Clogg replied:
NTPDATE will not adjust the time zone. It
will only accomplish the tiny adjustment to UTC time caused by the drift in
your system's hardware clock. There is no problem running both programs the
same night, as long as your DST adjustment has completed when you run the NTP
adjustment. Since the spring adjustment moves the clock forward, it will happen
immediately. The autumn adjustment can take a couple of hours or more, so it
calls for greater coordination.
Q: I have searched the archives and find
only a single reference to 'Device Class Limit' from 1996 that listed it at
450. Is that number still valid, and if not, what is the current limit?
A: Guy Paul replied:
You should be able to find this on Jazz at
http://jazz.external.hp.com/papers/limits/os_limits.html
Q: I'm setting Apache 1.3.4 up for on-line
access to reports. I used the fancy indexing feature for the directory index,
but when I display the index on a web page, it's truncating the links at about
22 characters. Does anyone know what can be done about this?
A: Mark Bixby replied:
You want to take a look at the NameWidth
option of the IndexOptions directive:
http://httpd.apache.org/docs/mod/mod_autoindex.html#indexoptions
Q: Is a patch required to install disks
greater than 9-gigabyte on MPEix 6.0? I am trying to install a Cheetah 36XL,
ST336705LW disk. I can format it, but it still shows as UNKNOWN when I do a
DSTAT ALL. This disk is to replace the ST39204LW, which was a 9-gigabyte
version.
A: Scott Swartzell and Craig Lalley replied:
It sounds like you haven't added the new
disk to a volume set in volutil. If you can see the disk in dstat all, you
probably don't need a patch.
Stan Sieler added:
The maximum size of a disk drive on MPE/iX
varies depending upon which release (or patches) of MPE/iX you're using. If I'm
in doubt, I check the IODFAULT.PUB.SYS file, and search for "GB" to
find what the largest supported disk might be. On our well-patched 6.0 system,
I see: '3.50" 73 GB FC Disk Device' which tells me that it had better
support a 73 GB disk.
By the way, "support" is a
slippery concept. For quite some time, MPE/iX hasn't done a proper job of supporting
larger disk drives. On 6.0 and later (perhaps earlier, too), MPE/iX would allow
you to mount/use a larger disk drive than DISCUTIL (an important offline
utility) would support.
Q: I recently had a problem with an Image
B-Tree index. A search like "Z@" (where Z was greater than all
present key values) took a long time to come back. Four minutes for the DBFIND!
On our test system I tried dropindex/addindex on a restored copy of the
database and that seems to have fixed the problem. This index is the built-in
IMAGE kind, not Omnidex or Superdex. There were 1.4 million entries in the
manual master dataset. So, my question is should we regularly re-index our
IMAGE B-Trees?
A: Jerry Fochtman replied:
This is one of those "it depends"
answers. If the dataset is fairly stable without a lot of add/delete, then it
could be re-built less frequently than the index for a set that experiences frequent
key add/change/delete activity. Keep in mind that KSAM/KSAMXL operates using a
'key' area and a 'data' area. The key area is dynamically maintained as a
balanced tree whenever values are added/removed. However, the data portion is a
'first-come, first-serve' approach so its possible that the data for logically
adjacent values are spread all over this area causing I/O to be extensive.
Especially when doing wild-card qualifications and it becomes necessary for the
data entries to be retrieved and evaluated against the request.
So the answer is 'Yes", it is a good
idea to re-built the indexes periodically. But at what frequency would have to
be determined on an individual basis simply because the dynamics of the data
and the affect it has on data locality are generally not easily predictable.
Q: Does anybody know offhand whether the MPE
intrinsic FPOINT works on Large Files
(i.e. > 4GB)?
A: Gavin Scott replied:
Of course it does. FPOINT takes a record
number, and Large Files can't have any more records than not-Large files, so
there's no issue with FPOINT or any of the rest of the Intrinsics for that
matter.
Q: The following does not appear to be a
problem. But for my own edification I would like to now why? We installed a
used HASS "Jamaica" A3312A drive enclosure with (2) 9GB FWD SCSI-2
drives. Both drive modules appear to be the same part number
"A5238A". I configured both drives in SYSGEN as "ST39236LC"
as per instructions from
HP support. However DSTAT ALL shows
different IDs. Why?
A: John Burke replied:
DSTAT is actually reading the ID information
off the drive. The difference between SYSGEN and DSTAT does not matter, though
many people alter SYSGEN after the fact with the correct ID so it is in sync
with reality. In configuring disk drives in SYSGEN, the only thing that really
matters is that you get the correct driver. On your system there are only two,
one for FW and one for SE. So choosing ANY FW drive would have worked.
Q: Does anyone know if it's possible to set
a var to the result of an finfo across a dsline connection? I want to retrieve
the status of a particular file on the remote system.
A: Donna Garverick replied:
You cannot do it directly. One way of doing
it, however, is to create a file equation on the remoted system that includes a
'nodespec' (read 'help all' for file) for the originating system. Then
echo ![finfo('myfile','myparm')]>*myfeq
That magic file equation will put the
information on your originating system. if it's a one-liner file, it's fairly
easy to get the input: (input myvar < finfofle). It involves some work, but
it's not too bad.
Q: Any help on the following error on a
message file would be appreciated.
+-F-I-L-E---I-N-F-O-R-M-A-T-I-O-N---D-I-S-P-L-A-Y+
! ERROR NUMBER: 71 RESIDUE: 8224 (WORDS) !
! BLOCK NUMBER: 538976288 NUMREC: 8224 !
+------------------------------------------------+
A: Per Ostberg, Jeff Woods and Stan Sieler
all replied:
"Too many files open."
Stan Siler added:
I've seen this a lot with COBOL programs
that open a file in utility code that then forgets to close the file. After
calling it about 1000 times (more or less), you'll start getting that error.
Q: I am trying to find information on the
Heat Dissipation for a Model 10 disk array, specifically the maximum load
BTU's. Where can I find this information?
A: Wirt Atmar replied:
You don't need to reference an HP site for
this information. It's available on the nameplate on the back of your device.
BTU's are a unit of work (or energy or heat; the concepts are all equivalent).
Kilowatts are a unit of power (work expended per unit time), thus
kilowatt-hours (units appropriate for the time integral of power) are
equivalent to BTU's.
If we presume that all of the electrical
energy flowing into an electrical device is dissipated as heat (which is not
only a good first-order approximation, it's also the most conservative
estimate), then the formula of interest is:
1 kilowatt-hour = 3413 BTU's
That is, a 100W (or 100VA) device will emit
at maximum 341 BTU's in one hour. HP however is conservative in its estimates
of power draws, so the real number will probably lie somewhere between 50 and
70% of such a number calculated off of the nameplate.
Q: I wanted to know how to check to see if a
file name fits the pattern IHOCS###, with ### being three numbers. Is there a
pattern matching function to do this?
A: John Burke replied:
Check out the CI function PMATCH.
Q: I have a port in HP3000 and I want to
know the application that is currently using that port. Is there any command
that can show me the applications accessing a particular port?
A: Kevin Miller and Jeff Kell replied:
:sockinfo.net.sys
Enter 'c' for 'call sockets'. Listeners are
shown in port order.