March 2005

net.digest tracks each month’s message traffic on the 3000-L mailing list and comp.sys.hp.mpe Internet newsgroup. Advice offered from the messages here comes without warranty, so be sure to test before you implement.

Edited by John Burke

Changes are afoot at HP. Again. We won’t have Carly Fiorina to kick around anymore, at least not directly. For the next year or more, however, she will be indirectly blamed for everything and anything that is wrong with HP, or anything else for that matter, including probably global warming. Such is the penance for failed CEOs. But do not weep for Carly. The $21 million she received to go away should soothe her hurt feelings. Recent reports suggest she is on the short list to be the next head of the World Bank. Since few people know what the World Bank does, this might just be the place for her.

There has also been some recent discussion that HP’s problems should be laid more at the feet of the Board of Directors. The theory goes that it is the Board that started HP down this path of commodity manufacturer and distributor some dozen years ago; that it is the Board that first lost touch with the HP way. From The HP Way, by David Packard: “Bill and I had no desire to see HP become a conglomerate, since, as I’ve already pointed out, more companies die from indigestion than starvation.”

The change at the top of HP engendered questions on 3000-L along the lines of, “Does this mean HP might now just Open Source MPE/iX?” No one outside or inside HP knows the answer for sure. While HP only has an interim CEO since the firing, with no obvious replacement looming, most HP programs and strategies will likely stay the same. Longer term, those who hope for a change in HP’s strategy to bury MPE should bear in mind that HP’s Board of Directors specifically said there was nothing wrong with the company’s overall strategy, just the execution of that strategy. Sorry about that.

I always like to hear from readers of net.digest and Hidden Value. If you spot something on 3000-L and would like me to elaborate on it, let me know. Reach me at john@burke-consulting.com.

Moving groups/accounts between volume sets

There is a suggested method for moving a group or account to a user volume from the MPEXL_SYSTEM_VOLUME_SET?” But the same answers apply when moving between user volume sets. Russ Smith offered the following step-by-step procedure:

a) !STORE @.@.account;DIRECTORY

b) !BULDACCT

BULDACCT: account%VSACCT=newvolname

c) !PURGEACCT account

d) !STREAM BULDJOB1

[#builds the account structure on the new volumeset]

e) !RESTORE @.@.account;VOLSET=newvolname;OLDDATE

[#will restore the files with their old create, access and modification dates intact]

f) !STREAM BULDJOB2

[#will reset any user UDCs]

The basic sequence is store-purge-restore. You can save yourself some time and potential aggravation if you use store-to-disk instead of tape in step number one. The situation becomes a little more complex if you split groups across volume sets, something that is often desirable. In this case, using the newacct/newgroup/etc UDC from HP’s Jazz will definitely help. HP’s Jeff Vance did a great job hiding all the idiosyncrasies of putting accounts and groups onto user volumes.

Moving data between arrays

One user had an N-Class system hooked up to an HP XP array and wanted to move everything to an HP VA array. Believe it or not, the only recommended way is a STORE, followed by a RESTORE. Ouch. Double ouch, since backup time was reported as 19 hours! After a lot of advice, the eventual plan decided on was to hook both the XP and VA to the N-Class box and migrate sets of data (groups) over time, because of minimal downtime considerations. No one was happy with this solution, however.

Walt McCullough of HP alluded to expensive third-party tools that move the data directly, but gave no references. Individual storage vendors, including HP, have server-less copy within an array, but apparently HP does not have a way to copy/move data between arrays. So, here’s a question : How can HP call itself a major storage player if it doesn’t have a way — other than slow, prone-to-error, tape — to move data among its own arrays?

Internal N-Class storage: When to use it, what for

Many sites have their N-Class systems connected to an XP, VA or some other array. But the N-Class systems have provision for several internal, non-RAID protected disk drives. The question is should you populate these drive bays, and if you do, what should you use them for. The consensus answer on 3000-L is you should have both internal bays filled. Use one for “Dump to Disk” memory dumps – large memory systems take roughly forever to dump to tape.

The second disk should have a bootable copy of MPE for diagnostic purposes and for that nightmare time when you lose connectivity to your array. As for using any of the internal drives for part of the system volume set, don’t! There is no protection, and even if you have the other disks of a multi-disk system volume set on your array, if the internal drive fails, you are toast.

Quick Cuts

• Do you want to know when a particular account or group was created? LISTACCT and LISTGROUP are no help. But “listfile /ACCOUNTNAME,3” for the account or “listfile /ACCOUNTNAME /GROUPNAME,3” for the group tell all. And then some.

• The number of sectors reported by the REPORT command for a group or groups is sometimes inaccurate, sometimes very inaccurate. Running the program FSCHECK.MPEXL.TELESUP and issuing the SYNCACCOUNTING command will fix this problem.

• In case you were wondering, despite many requests for the enhancement, TurboStore will NOT append store sets to tape. Well, it might if you use the proper incantations, but it is unsupported and highly dangerous because under certain circumstances you could overwrite a previous backup without knowing.

• Speaking of things you cannot do that you might like to do, the ALLOW command is not persistent across sign-ons unless you use the extremely dangerous “ALLOW @.@; commands” version. This is another example of an enhancement that has been requested for years, but now will never happen. Fortunately, there are a number of options, for sale and free (MPEX, CSL, etc.).

• CI integer variables are signed 32-bit entities. So be careful if you are doing some wild arithmetic in your CI scripts.

• Here is a little trick when using Apache’s indexing (for example to keep track of documentation) to index file displays. You can override the default ascending sort by name by appending “?N=D” to the url. Instructions on changing Apache’s default behavior are available at httpd.apache.org/docs-2.0/mod/mod_autoindex.html#indexorderdefault.

• If you are trying to program VPlus applications and are interested in working examples programmed in your favorite language, look in the group HP32209.HPPL89 (which should be on every FOS tape). This group contains source code for the ENTRY program in a variety of languages including COBOL, Fortran, Basic and Pascal.

• To see the firmware (aka PDC) Revision of a system (CPU): Run cstm, and at the cstm > prompt, type ‘map’ and note the Dev Num of a CPU and then type ‘sel dev DEV_NUM’ (e.g., ‘sel dev 41’) and then type ‘info’ and then type ‘il’ and look at the output for the ‘PDC Firmware Revision’. Easy, huh? Thanks to Guy Paul of HP for this tip.

• SPFXFER will allow you to write to disk (undocumented “feature/bug”). But don’t do it, because SPFXFER cannot read the disk file it creates! Doing this could lead to a big oops.

• While it would certainly be a nice to have, MPE/iX CI scripts have no provision for inline comments. Sorry, don’t even bother trying.


Copyright The 3000 NewsWire. All rights reserved.