Post the most significant problem your have solved on HP-UX and win up to 10 points....

[pap]

That will help us a lot guys...thanks so much for your contribution in advance.

-pap

[John Poff]

Hi Pap,

Sorry. I'd love to help you but the offer of a ten point bounty would earn me the label of a points whore by many of the high and mighty members of the forum, and God knows I wouldn't want that!

JP

[John Bolene]

Installing expect and writing an expect script to auto update my password on 100 machines monthly.

This used to take me most of one very boring day to log on to all of them.

[Sachin Patel]

PAP,
I did post in your thread is the most significant problem solve of my ...... Where is my 10 point,,,,???

kidding don't even try it.
Many people will going to slam you for this bounty.

But it is good interesting thread.
If you promise to give me only one point I will post back.

Sachin

[Sachin Patel]

Pap,
Pelople are going to slam you for this bounty.

My most significant problem solve is to reply here.
Where is my 10 point????
just kidding

If you will give me less then 3 point then I will reply in this thread again.

Sachin

[Sachin Patel]

Pap,
Pelople are going to slam you for this bounty.

My most significant problem solve is to reply here.
Where is my 10 point????
just kidding

If you will give me less then 3 point then I will reply in this thread again.

Sachin

[Donny Jekels]

Is this how you get more than 1000 points?

hmmm.... pretty lame indeed

[Ramkumar Devanathan · Accepted solution]

Here's a real good tech thread. John and Sachin, don't throw out the 'original techie itrc' baby with the '10 point congratulatory thread' bathwater...

So pap, *my* most significant problem solved - is no great big shakes in terms of knowledge coz every hp-ux person has done this at some point of time or the other - but here goes anyway for posterity or whatever they call it -

When you are installing a patchset containing especially kernel patches, and the machine thereafter is to undergo a reboot, You better make a copy of your /stand/vmunix say as /stand/vmunix.golden and then install the patchset. This way even if the machine does not boot up after the patch install you would be able to go into single user mode using the following set of commands -

- interrupt the boot process press any key (don't ask "where the hell's the 'any' key?")
- type "bo pri"
- Interact with IPL? --> Y
- type "hpux -is /stand/vmunix.golden"

If you do have a backup kernel, this command will boot the system from it and take you to single user mode. Now you can build yourself a new kernel or copy /stand/vmunix.golden to /stand/vmunix.

FWIW.

- ramd.

[pap]

Hi John,
I can understand your concern about the title of the thread. However, it is kept for FUN as you know....!!!
Nothing to become serious about that.

Sachin,
To get 10 points , you need to show something ..otherwise no one can help you.with points ...:))..

I know this thread will going to help us a big time , I just want that all forum members take part in and make this thing successful. I do not think there is something wrong if I give 10 Points to the member who has solved great problem....Anybody is against this???..please let me know....if this is the case....I can withdraw my post and will not commit the kind of mistake again....


-pap

[Steven E. Protter]

Most significant problem.

PHKL_18543 LITS (Line in the Sand)

I installed it.

I installed it again.

I suffered months of pain and suffering recovering from it.

The symptoms:
1) System panic at boot
2) Could not generate a new kernel
3) Lots of corrupted patchees.

1) Did not realize the problem until Ignite tapes had all rotated through, so I had to boot the box and go to vmunix.prev Still, not having solve

Fixed symptom 2 by restoring the sytem file from a very old backup and reinstalling all of the drivers installed after that point. This took a great deal of time.

3) Patch structure was a mess. Dozens of corrupted patches. To avoid HP's solution of total OS install on an old but production box, for the next few months in maintenance windows, I reinstalled the patches that were damaged with -x reinstall=true

Finally I ended up with a production system with no reinstall and less than 8 hours of prime time down time.

As to the earlier commentary, I think everyone should simmer down and be a little more supportive of one and other.

Self congratulations threads are somewhat over the top but, you don't get points out of it so maybe its not so bad.

http://www.isnamerica.com
SEP

[Elif Gius]

We had a problem on one of our L-class server.
The system was patched with a patch bundle and must be rebooted for the kernel patches. But then it was rebooting all the time. We could look on the console what he is doing during the startup process - all looked fine. We could even log in , but then the system was rebooted again...
So I tried following:
I interrupted the boot, went into singlue user mode ... so far so good ;-)

then I went into init 2, nothing happend...then
I had some time for looking the log files, what was wrong..I looked the syslog - but all was ok. So I had a look at the rc.log, because I wanted to know at which point of the startup procedure the system went down.
There were only a few startup procedures for init 3. So I didn't have much problem to find the causing - the Apache startup script.
I moved the startup script from the init-Procedure. And then I rebooted the system again.
It looked find - and the system remained up.
So it was the Apache Startscript!!! WHY ???
I'm not really sure. The problem was that this apache was a oracle apache...which was installed too with the oracle instance. But then the oracle instance was removed, but the startup script remained. He tried to start apache with his config files which already didn't exist anymore. And then the system rebooted.

That's all. But now I now in fact how I can interrupt the system from booting again,and again, and again, and again, and again ...

[V. V. Ravi Kumar_1]

Hi,
"Win up to 10 points" caused all this... any ways this is a significant problem, which i had created my self.

We have two L-class machines in a cluster. We have configured MCSG package to access one VG on the common RAID. One day we got a problem on the secondary node to access the VG, we decided to re-import the VG from the primary node.
I did vgexport on secondary node.

Created directory for VG and group file again.

Then on primary node i should have done vgexport with -p option to create map file. But by mistake i did just vgexport. At that time the VG was not active on Primary node.
So i lost the connection to VG from Primary node also. Then i understood that iam in trouble. I imported the VG with PV Path, i got all the LV's but with
lvol
lvol2
lvol3 etc., original names have gone.

Fortunately we had a map file backed on some tape, where the VG id was different, i retrieved that file.

I created the map file with the present VG setup and replaced the VG id in the old map file with the new VG id.

Exported the VG again from the Primary node, imported with the old map file, i got everything right. It was big relief.

From then onward i included the follwoing line in the crontab to have backup of map files of all VG's on every Friday.

0 9 * * 5 vgexport -p -s -v -m /etc/map/vgbm.map


Regards

[Michael Tully]

I don't want to upset people, so perhaps your wording of your posting was not good.
However we are in this forum to assist others and to share experiences and knowledge. So I am going to share an experience, not so much in the way of knowledge.

This was quite simple really, the first day I walked into my current job, there was a networking problem with lost packets, slowness and generally loads of complaints.
The first thing I did (after putting my bag down) was determine the communication of the LAN card to the network. Bingo the card was running as half-duplex, instead of full-duplex.
I fixed this by issuing a 'lanadmin -X 100FD 0' which fixed the problem.

In the end everyone was happy including me, as I was able to go and get some coffee.... must needed on your first day ....

Cheers
Michael
(no points please ...)

[Pete Randall]

I don't know if this was the most significant, but it was certainly the most vexing. I finally ended up figuring the infamous "at job enema" problem myself, after a little help from my fellow forumers:

http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x1cb03a7b3682d611abdb0090277a778c,00.html

Of course, as usual, it turned out that it was something that I'd done: a recently installed patch started the problem.


Pete

[Pete Randall]

Darn! I forgot to use my new signature:


Pete, The Point Whore

[Umapathy S]

my 2 cents.

The story goes like this. I was trying to add new user through SAM and it refused me replying "Permission denied". I was perplexed. Then with 2 days of digging in the problem among other works and with the help of our gurus here, I was able to solve it.

This is the second thread which has the solution.
http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0xb40b6049dbb6d611abdb0090277a778c,00.html

This is first one which I raised when got the error.
http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x05046049dbb6d611abdb0090277a778c,00.html

My experience is if you systematically study and track the problem with the help of experts you can crack it.

thanks
Umapathy

[Steven E. Protter]

Can someone please explain to me what all this "point whore" talk is all about?

Apologize for the off topic note, but there are people with religious and other sensitivities on these forums. I personally find such talk distracting from the purposes of these forums.

I could post a dozen significant problems solved.

I just avoided getting bit in the tush by the dreaded pax patch.

The pax patch enables pax to pack files up to 8 Gigabytes into an Ignite golden image.

BUT, pax_iux the utility that unpacks the image after the image is pushed can't handle files bigger than 2 G.

So what happens if you miss this is your image stops loading and you have no idea why.

I noticed my golden image last night that I was loading a downed production system off of was a little large.

I looked up the itrc case, swlisted for the patch and sure enough the darned thing was back in again(it must have gone in as a dependency).

Had to stop the Ignite load, redo the image and just started rebuilding the production server.

Call me anything you want(even a point whore), but don't forget to call me to dinner.

Shabbat Shalom
Have a good weekend.

http://www.isnamerica.com
Steven E. Protter

[Pete Randall]

Sorry Steven, didn't mean to offend anyone - I'm just being up front about it after all the furor yesterday. See
http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x022718276953d61190040090279cd0f9,00.html

In particular, look for JRF's comments and Jeff's response. Frankly, I took offense, and this is how I choose to deal with it - by being up front about it. If that's what they want to call me, then I might just as well be proud of it.


Pete, the Points Monger

[Mark Greene_1]

I'm not sure how significant this is, but it was a challenge. We have a multi-port serial board on our L2000 that we've attached 2 modems to. I config'd one of the open ports to be a "back-up" access point for the console in the event we lose the regular console serial port. I was eventually reduced to trial-and-error setting-up /etc/gettydefs to get it to play correctly with the HP terminal. Here's what I ultimately arrived at:

serialconsole # B9600 HUPCL IGNPAR ICRNL IXON OPOST ONLCR CS8 CREAD ISIG ICANON ECHO ECHOK IXANY CLOCAL
# B9600 HUPCL IGNPAR ICRNL IXON OPOST ONLCR CS8 CREAD ISIG ICANON ECHO ECHOK IXANY CLOCAL
# login: #serialconsole


mark

[Steven E. Protter]

No problem Pete.

I actually participated in the discussion, but clearly didn't understand it.

My posts were part of the dup problem, and Berlene Herren kindly removed all my unpointed duplicate posts. She's working an itrc support case for me right now and doing a fabulous job.

Good luck.

http://www.isnamerica.com
Call me anything you want, but don't forget to call me to dinner.
Steven E. Protter

[Brian M Rawlings]

Not exactly steller HP-UX fixing, but...

I spent a month one week trying to load HP-UX 11 and patches, etc, onto a used R380. HP-UX appeared to load OK, but between patches and HP App loading, I had the darndest series of flakey problems that prevented me from getting even the general release patches loaded.

After a long series of swinstall sessions, which would appear to go fine, except that somehow the dependencies were never met (missing files, cannot continue, repeat load, check files, they're all there, back to swinstall to finish, and, poof, missing files...) Log entries showed no errors.

After working through three CD sets with unending, vague, never-the-same-way-twice issues, I finally decided that HP had to send us a replacement internal DVD reader for the used R380 (warranty). The customer said that it couldn't be that, and began to question whether I could load HP-UX at all.

The new DVD drive came in, and there were no more problems -- HP-UX loaded fine, patches and apps loaded fine, customer apps and data base loaded fine... Man, what a week. It's not much use getting lots of extra practice loading the OS if you begin to question your sanity and abilities in the process. The customer never did acknowledge that I'd been right about the DVD, and had been doing it right all along.

The clincher was that, once the patches and OS apps (mirrordisk, glance, etc) were loaded, the R380 could not see the Autoraid 12H. Logs showed lots of SCSI errors (lbolt times were dozens per second). Terminators, cables, alternating 12H controller cards and FWD HBAs in the R380, nothing helped.

pulled the two A4107A FWD HBAs out of the R390 to check any switch settings, and, you guessed it, no little yellow Terminator packs in the sockets. Scrounged some and installed them, and everything worked as expected.

Lessons learned: if you get used equipment, even from HP, assume that everything's bad and messed up. Don't assume even the slightest testing has gone on. "Refurbed to factory specs" is a marketing term, not a technical description. And, if you get something from a broker, it may be worse (but you can usually talk to the lab guy that put it together, and find out what testing HE did). And, I guess, the same crap could happen from new gear as well, although I've never had this much pain from new stuff.

Later...

--bmr

[Anil C. Sedha]

Attaching 2 posts :

This post is important, cause other people did not think much over how much problem does it give for upgrading 11.0 to 11i (no issues) if we take simple care and it's very simple. - Anil

November 14, 2002 21:50 PM GMT [ 8 pts ]

--------------------------------------------------------------------------------
Hi Sean,

I also upgraded my OS to 11i and there were no big issues, apart from the new compiler version that you will have to install.

One very major concern, the root fs and /stand fs require specific filesystem size. You may have to increase their size, but it has to be done with care as they are sequential and hence you have to remove primary swap and then do it. If you know the steps, fine else let me know


You will have to upgrade ServiceGuard, C++ Compiler, JDK. Omniback - I am not familiar with so i can't say.

Rest of the stuff works fine. On the time part, the OS upgrade takes anywhere around 2-4 hours depending on your system. Rest of the applications are fast.

Regards,
Anil

A 10 points answer
http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x5e2e06350fe2d61190050090279cd0f9,00.html



Enjoy
Anil

[Anil C. Sedha]

Attaching 2 posts :

This post is important, cause other people did not think much over how much problem does it give for upgrading 11.0 to 11i (no issues) if we take simple care and it's very simple. - Anil

November 14, 2002 21:50 PM GMT [ 8 pts ]

--------------------------------------------------------------------------------
Hi Sean,

I also upgraded my OS to 11i and there were no big issues, apart from the new compiler version that you will have to install.

One very major concern, the root fs and /stand fs require specific filesystem size. You may have to increase their size, but it has to be done with care as they are sequential and hence you have to remove primary swap and then do it. If you know the steps, fine else let me know


You will have to upgrade ServiceGuard, C++ Compiler, JDK. Omniback - I am not familiar with so i can't say.

Rest of the stuff works fine. On the time part, the OS upgrade takes anywhere around 2-4 hours depending on your system. Rest of the applications are fast.
-------------------------------------

Another one
http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x5e2e06350fe2d61190050090279cd0f9,00.html



Enjoy
Anil

[W.C. Epperson]

I think it was the update from hp-ux 8 to hp-ux 9. After updating and configuring all my stuff, the system went into a rolling boot-panic-boot. Managed to alternate boot into single user mode, and spent a couple of days picking things apart. Turned out HP had shipped the new version with what appeared to be the drivers for HP-IB disks, but the disks weren't supported and the drivers were totally incompatible with the system.

Then there was the "upgrade" from SNA Classic to SNAPlus. I almost needed detoxing after that one.

But if you want the fastest I ever fixed a huge screwup on an HP box, it was under MPE on a 3000. I naively ran a program called "initrin" to find out what it did. It printed a console message that said "Hi. I zero out the Global Resource Initialization Number table and do a Halt-B on the system". And before you could read the whole thing, it did. Had to manually reload the GRIN table to get HPOffice working again. With the computer room phone ringing off the hook the whole time.

[Roger Crettol]

Breaking tlfileset's neck ...

Not the recommended approach, but I got tired of clocking 'install cleanup' times on my long-breathed D350.

I've seen forum posts dealing with this problem but couldn't find them again. Deinstalled the UPG-TLINK only to be sorry about the missing links, re-installed UPG-TLINK, and digged in the grease ...

Perl's way of munching thru the swlist output brings joy to that old system, at last.

Net worth : +5 points (brave an silly) and -5 points (dont' touch HP's original code).