We use a backup product that provides software data compression. As an experiment I enabled hardware compression on our 959's DDS2 drive and performed a full backup. There does not seem to be any significant decrease in backup time. Why is that, and how can I speed up my backup?
Mark Klein replies:
There are a couple of things to consider when examining the trade-off between hardware and software compression. When software compression is on, less data is passed down the channel to the drive at the expense of the CPU. When software compression is off and hardware compression is on, more data is passed down the channel, but nearly the same amount is finally put on tape.
Now, depending on your environment, if you're I/O or CPU bound, the results of enabling hardware or software compression can significantly change performance. If you are neither I/O nor CPU bound, the effects of such a change between hardware and software compression should be minimal.
You effectively end up with the following matrix:
I/O Bound: Since you already have I/O bandwidth problems, you should
minimize the
amount of data sent to the backup device in this case. You want software
compression
with no hardware compression on the target device. A word of warning here:
using high
software compression with hardware compression enabled may actually cause
the data to
expand over what it would have been without any compression whatsoever. Do
not do it!
CPU Bound: Since software compression can require heavy doses of CPU
depending on
the method chosen, you should pump the data down to the device for
compression. This
keeps the CPU available for other tasks at the expense of greater I/O.
Both I/O and CPU: This one gets too tough to really discuss here. There are a lot of trade offs and switches with which some backup products can be tuned in such an environment. Each case tends to be different. Your software support people are best equipped to discuss this with you.
Neither: In the case where you're neither CPU nor I/O bound, the speed of the device will become the limiting factor, and you will not see an appreciable difference between software or hardware compression.
How can I automatically put a DAT tape back online without having to
manually take the
tape out and push it back into the drive?
Siegfried Gladitsch of HP replies:
Starting with MPE/iX 5.5 you can use DEVCTRL.MPEXL.TELESUP:
DEVCTRL.MPEXL.TELESUP ldev load=online
John Pearce replies:
If you have MPEX, check out the DEVCONTROL command.
Keven Miller replies:
The Interex CSL has a contribution by name of ONLINE. My index says Sanfran/93. Both a PM and NON-PM version.
Stan Sieler replies:
Go to the URL www.allegro.com/software/ and click on "Allegro programs", and look for ONLINE:. Or if you know how to use the LZW compression program, the direct URL for the LZW package of the ONLINE program is: http://www.allegro.com/s oftware/stuff/ONLINE.LZW
No question, just a holiday gift. Doug Werth writes:
I ran across something the other day that I have never seen in any of
the sites I have dealt
with, nor have I seen or heard it discussed. It is a file called
NMSTART.PUB.SYS and it
sets controls for NMMGR. A sample file would look like this:
:print nmstartThis file will accomplish two things. First, it won't allow anyone
to modify the NMCONFIG
file without the password. Remember the 'write access password' at the
bottom of the
opening screen in NMMGR? Second, it will automatically validate dts and
netxport upon
exiting NMMGR (and then only if necessary).
password maryxmas
exitvalidate dts on
exitvalidate netxport on
I know that MPE/iX 5.5 has an HP variable that gives the IP address of the user. But I'm on MPE/iX 5.0. Is there anything I can use?
Duane Percox replies:
Check out http://www.qss.com/vtverify.zip. It's the location of VTVERIFY, a simple COBOL program that calls NSINFO to return the IP address of the user executing the program and uses a simple file as a filter table to allow you to screen out users by IP address. The zip file contains the source, the binary, notes and a sample filter table file. This is freeware for your own noncommercial use.
For the second time in less than one month I got paged by our backup job
with an error.
In each case the problem was with a different tape drive but in both cases
a 120-meter
DDS2 tape was involved. The error messages on the console were:
MPE XL Status: Subsystem 113, error -44These were the first 'media' errors since we got these systems
almost two years ago. We
started using the DDS2 tapes sparingly in the last few months. Having two
errors in a
month makes me wonder if there may be something inherently wrong with the
120-meter
DDS2 tapes or just with our batch. How can I check them?
The media on device 'xx' is bad
Mike Hornsby replies:
In SYSDIAG you will find a utility called SCSIDDS. If you are on MPE/iX 4.0
or later, you
will have to call HP for the password.
run scsidds;section=50;ldev=#This utility will allow you to display the usage and error logs for
the media. This will
probably help to determine if it is the media or the drive. I have
repeatedly suggested that
STORE have a TAPELOG parameter to display these logs after each media
usage. [Editor's
note: the new STATISTICS option of MPE/iX 5.5 STORE provides some of this
information]
Does MPE/iX support multiple IP addresses on a single LAN card? In HP-UX this is supported and is called "ip aliasing" and is set up with a program called ifalias.
Chris Bartram replies:
Sorry, no. Those of us hosting web servers can appreciate the usefulness of IP aliasing. Our Linux servers nicely alias any number of IP addresses, allowing one physical system to "host" many different organizations.
To which Mark Bixby notes:
The HTTP/1.1 version of the protocol allows a Web client to pass the hostname to the web server, so virtual web serving will no longer require IP aliasing once all clients support HTTP/1.1.
How can I get MPE/iX 5.5 patches online?
Joe Geiser replies:
MPE/iX patches exist at us-support.external.hp.com
either via www or anonymous FTP.
Fire up FTP (seems to be the best way), and:
open us-support.external.hp.comOne little gotcha. There are a couple of patches that are over 20Mb
in size. If using
FTP.ARPA.SYS, you will need to add "buildparms" to get around the default
file size. It
would be nice if HP put the size in sectors somewhere so we know how to set
buildparms. But the good news is that you can get them.
user anonymous
pass (e-mail address)
cd mpe-ix_patches/c.55.00
get (patchfilename)
I am having trouble with MPE/iX patches downloaded from the web site (us
support.external.hp.com). The unpacking process seems to fail in MOVER.
HP's Wray Smallwood replies:
MOVER is used in the patch download process from the Web. Recent problems reported with MPE/iX 5.5's version of MOVER.PRVXL.TELESUP have resulted in the following discoveries.
There are two different versions of MOVER A.00.02, with identical headers. The version submitted to 5.5 has an EOF of 651 while the version obtained from jazz.external.hp.com or patch MPEHX12C has an EOF of 635.
CSY has been using the version from Jazz to compress files. This worked fine for 5.0 patches because the version obtained from Jazz and the MPEHX12C patch are the same. However, in 5.5 the submitted MOVER does not appear to decompress all the files compressed with the jazz version. Errors occur showing an incompatibility between the two, and files are missing. This is a contributed program that CSY will endeavor to fix.
CSY Action: Of course we will attempt to ensure that the 5.5 version does work. We will put out MOVER A.00.03 which is essentially the 5.5 version with a new header. At that time we will also redo all download files to make them A.00.03 versions. We will also redo UNPACKP, although the old UNPACKP should still work, so that it specifies MOVER A.00.03. We will create an orderable patch which will make it into any future PowerPatch tape. The patch ID will be MPEJXF8A for 5.0 and MPEJXF8B for 5.5. A.00.03 will also be put on jazz.external.hp.com. And, it will be submitted to the next opsys. An announcement will be made when you are to start using A.00.03 on newly downloaded files.
Your Action: On 5.5, preserve your copy of MOVER A.00.02 (EOF 635) by restoring it from a 5.0 tape after installation. We will not compress any files with the 5.5 version. However, you might want to keep it in case you obtain files from another source that uses it. When we have A.00.03 available, it will be available as a patch or a download on jazz. Obtain UNPACKP A.00.02 from the Web at the same time.
I am trying to do a DBLOCK, DBPUT, DBUNLOCK in sequence in a program. However, the DBLOCK fails, saying a lock is already in effect. Then the DBPUT fails saying there is no lock in effect. I looked in DBUTIL to see if any locks were held immediately before and after the call to DBLOCK, and no locks are in effect. How can I fix this?
Jim Phillips replies:
I have seen DBLOCK fail when the program does not have MR and locks another file object, such as a flat file or a KSAM file. In such a case, DBLOCK fails but DBUTIL shows no locks for the database because the lock is held in the MPE file system, not IMAGE.
What are the requirements for setting up a remote DTC using the routable
DTC protocols
(AFCP)?
Mike Yawn and John Kramer reply:
An overview is in the MPE/iX 5.0 Communicator article "DTC/TIO Enhancements in
MPE/iX 5.0", pages 7-21. Detailed instructions are contained in "Using
OpenView DTC
Manager", which comes with HP OpenView DTCMGR. A super abbreviated version: