DECWindows does not start

[C.Eichert]

On an Alpha Server 1200 with OVMS 7.2-1 suddenly DEC Windows does not start after booting.
Means the process DECW$SERVER_0 is started but not DTLOGIN and DTGREET. The screen is flickering with black/white diamonds, the mouse pointer is visible and can be moved.
Stopping and restarting DECWindows does not help. The upgrade VMS721_GRAPHICS V5.0 hasn´t solved the problem.
For testing I swapped the graphic adapter - but the same.
Does anybody knows the Problem?
Thanks
Christo

[Ian Miller.]

What else changed on this system?

Can you post the result of SHOW CONFIG command at the console (best as a txt file attachment).

[Wim Van den Wyngaert]

I would check :

1)sys$manager:*decw*.log
2)accounting for process failures
3)audit
4)operator log file

(as for most problems)

Wim

[C.Eichert]

Hello Wim,

attached are the files DECW$SERVER_0_ERROR.LOG and OPERATOR.LOG

I can not see some relating messages?

Christoph

[Wim Van den Wyngaert]

Is "Namespace Name invalid or not in cache" a normal message ? Could it be that decw is not starting because of decnet not set up correctly ?

Did you also check device errors (sh err) ?

Did accounting report process failures ?

Wim

[Volker Halle]

Christoph,

did you watch console output during boot ?

You may consider setting STARTUP_P2 = "D" to automatically redirect SYS$OUTPUT to SYS$SYSTEM:STARTUP.LOG during STARTUP - then check that file for unusual error messages.

What's the state of OPA0 (SHOW DEV/FULL OPA0) ?

Volker.

[Wim Van den Wyngaert]

D can be dangerous is you use sysman to do the startup.

Extract of my old mail :

To receive a dump of the boot messages in a file, I had put the system parameter STARTUP_P2 on D on all arb stations.

Undocumented is however, that this D is passed to all procedures declared in
sysman (sysman startup show files *) WHEN no parameters are declared in sysman for the procedure.
Thus lat$startup was executed with P2="D" which was interpreted as being the node name.
Since many stations received the same name (D), the communication with the printer went wrong.

Wim

[C.Eichert]

Hello Wim,

- I do not know the relationship of DECWindows and DECNET. On an other server we are using DECWindows without having DECNET configured at all.

- sh err reports "no device errors found"

- what do you mean with accounting

Christoph

[Wim Van den Wyngaert]

Chris,

$ acc/fu/sin=xxx [accountng:]/out=toto.txt

and post this file.

If the DT processes failed, there should be a record of it WITH the reason why it failed.

Also check if "show acc" gives a line with PROCESS. If not, "set acc/ena=process".

Wim

[C.Eichert]

Hello Volker,

I set Startup_P2 to "D", however to reboot I have to wait unit tomorrow.

sh dev/full opa0

Terminal OPA0:, device type unknown, is online, enabled as operator terminal, record-oriented device, carriage control.

Error count 0
Operations completed 185
Owner process ""
Owner UIC [SYSTEM]
Owner process ID 00000000
Dev Prot S:RWPL,O:RWPL,G,W
Reference count 0
Default buffer size 80

Christoph

[C.Eichert]

Hello Wim,

attached the output of acc/full/... at boot time.

show acc reports among others: PROCESS any process termination

Christoph

[Volker Halle]

Christoph,

there are a couple of detached processes exiting with %LIB-F-ACTIMAGE - that looks suspicious.

Your OPA0: looks o.k.

When did this problem start ? Can you look at the accounting records for the last boot, when it still worked ? Do you also see those %LIB-F-ACTIMAGE errors (if so, it may be a different problem) on the last good boot.

Volker.

[Wim Van den Wyngaert]

Chris,

I agree with Volker. And decw is the last thing started during boot and the last message before the interactive login in the accounting file is a process terminated in error.

Some logical must be missing pointing to the executables. Or may be the exe's were removed by some accident / upgrade ?

Wim@home

[Wim Van den Wyngaert]

May be transports or cde/decw are misconfigured. Check for changed decw files on your system disk.

Also read http://groups-beta.google.com/groups?as_q=actimage+decwindows&num=100
&scoring=r&hl=en&as_epq=&as_oq=&as_eq=&as_ugroup=&as_usubject=&as_uauthors=&lr
=&as_qdr=&as_drrb=b&as_mind=1&as_minm=1&as_miny=2000&as_maxd=9&as_maxm=5
&as_maxy=2005&safe=off

"Edited to reduce width of screen display. The above should all be one long URL"

One of them might be your case.

Wim

[Volker Halle]

Christoph,

if the %LIB-F-ACTIMAGE actually comes from the DECwindows processes, you could try to just manually start DTGREET to see the error message and the file failing file name on your screen:

$ run SYS$SYSDEVICE:DTGREET.EXE
(line wrapped !)

Volker.

[Wim Van den Wyngaert]

Volker : since the image wasn't found, I don't think it will give the error message.

I've seen simular problems when decw was badly setup but I don't recall any details except that the cde$system_defaults logical wasn't defined while cde$startup was executed.

Wim

[Volker Halle]

Wim,

the %LIB-F-ACTIMAGE error should have come from trying to activate a shared library. It would have been %DCL-W-ACTIMAGE, if the main image could not be found.

Volker.

[Wim Van den Wyngaert]

Oeps !

[C.Eichert]

Hello Volker,

attached please find startup.log. So far I can not see some relating messages?

Christoph

[C.Eichert]

Hello Ian,

attached is show config at console prompt, yesterday I was unable to shutdown the server.

Christoph

[Volker Halle · Accepted solution]

Christoph,

what's the process id of DECW$SERVER_0 in your running system ? My guess would be 0000011A (from the last %RUN-S-PROC_ID in your STARTUP.LOG).

The accounting file posted earlier shows %LIB-F-ACTIMAGE errors for process 0000011B

In my system, the next processes after DECW$SERVER_0 are DTLOGIN and DTGREET (their process IDs are 1 resp. 2 higher).

Try to manually run DTGREET to see, which file cannot be activated.

Volker.

[C.Eichert]

Hello Volker,

your guess is ríght. The process id of DECW$SERVER_0 is 0000011A. The %LIB-F-ACTIMAGE error for process 0000011B is the next one that is DTGREET. Manually starting DTGREET reports: error activating image, too. After checking the bin Directory I found that it is completly empty. What is the best way to reinstall DECWindows from the installation CD?

Christo

[Volker Halle]

Christoph,

in general, a product can be removed with PRODUCT REMOVE and installed with PRODUCT INSTALL.

Removing or re-installing DWMOTIF is not something, I've ever tried. A good backup of your system disk would be a pre-requisite... Ideally, I would try this on a copy of the system disk on a test system:

$ PRODUCT REMOVE DWMOTIF
$ PRODUCT INSTALL DWMOTIF/SOURCE=vms-cd:

Good luck,

Volker.

[C.Eichert]

Hello Volker,

I simply copied the complete directory [.system...]*.* from the installation CD to the same directory of the server. DECWindows is working again as before. Thank you for the help.

Christoph

[Wim Van den Wyngaert]

C.

Just hope that the next system manager will find out what you did. You should always try to install using product or vmsinstall.

Wim