Conference csc32::consolemanager

879.0. "1.6 ECO 1 problem - Console Ctrl 02 inaccessible" by CERN::HOBBS (Protest the French bomb - kick a poodle today!) Thu Jul 20 1995 08:08

From:	VBORMC::"[email protected]" "Tim Whibley CN/DCI/GS - tel 4849"
19-JUL-1995 16:55:45.56
To:	<cern::hobbs>
CC:	
Subj:	PCM problems - long/urgent/wide

CW,

I installed the ECO1 patch this morning and after this, PCM gets restarted.

All the console status messages (cons stat/all) report no problem but I cannot
access consoles attached to Console Ctrl 02. The console reconfig hangs and I 
can no longer do a restart of PCM.

Inevitably, a reboot is required to clean up. Following a period of stability
prior to the installation of the SSB kit,

I have now seen the following problems...


12-Jul-95 09:00  installed SSB kit
17-Jul-95 16:00  archive procedure never completed and unable to reconfigure.
		 Tried to restart but Console Ctrl 01 had to be stopped
		 proc/id - system eventually blocked and reboot required,
	* high priority*

18-Jul-95 11:00  systems attached to Console Ctrl 02 inaccessible - reconfig
		 hung - reboot
	* high priority *

18-Jul-95 14:30  added new user and tried to reconfigure - never completed -
		 stopped all controller processes and finally restarted PCM
		 but one controller would not restart - reboot
	* high priority *

19-Jul-95 09:00  installed ECO 1

19-Jul-95 14:00  all systems attached to Console Ctrl 02 inaccessible although
		 cons status is OK- reconfig hung - restart PCM - console ctrl
		 02 only console process left in system as other controllers
		 abort with "Read error on Local socket CONSOLE_CTRL". stop
		 proc/id for console ctrl 02 and attempt to restart PCM
		 again - no controller processes appear - reboot
	* high priority *

I have currently 59 consoles attached via PCM and expect this number to reach
100 in the near future. PCM is a key component in our hetrogeneous system
management environment, connecting Suns,HPs,RS6000s etc. It is critical that
this software is robust and system managers expect access to their consoles
24 hours a day and rebooting the system is not a viable solution.

I am prepared to do any amount of testing necessary to get this software
working. Questions I would like to ask are ...

   1) are there any plans to use the primary/failover host ?

   2) would a move to an Alpha Digital UNIX platform (I currently run on an
      80 Mb VAXstation 4000/60) reduce the amount of downtime that I
      experience with PCM ?

% ====== Internet headers and postmarks (see DECWRL::GATEWAY.DOC) ======
% Received: from ns1.digital.fr by vbormc.vbo.dec.com (5.65-jep/rmc-umc-03jul95)
id AA26888; Wed, 19 Jul 1995 16:29:33 +020
% Received: by inet-gw-1.digital.fr (5.65/VBE-jep-20mar95) id AA01845; Wed, 19
Jul 95 16:56:24 +020
% Received: from VXCERN.DECnet MAIL11D_V3 by dxmint.cern.ch id AA20164; Wed, 19
Jul 1995 16:55:36 +020
% Date: Wed, 19 Jul 1995 16:55:36 +0200
% Message-Id: <[email protected]>
% From: [email protected] (Tim Whibley CN/DCI/GS - tel 4849)
% X-Vms-To: vxcern::cwhobbs
% X-Vms-Cc: TIM
% Subject: PCM problems - long/urgent/wide
% X-Mail11-Ostype: VAX/VMS
% Apparently-To: <cern::hobbs>

cw,

  Could you please have your customer officially escalate these
  problems through IPMT so they get tracked correctly during the
  transition of PCM support to Israel.

Cheers,
Phil

    Any comments about the question
    
    	Is Digital UNIX more stable than OpenVMS?
    
    -cw

    >    Any comments about the question
    
    >        Is Digital UNIX more stable than OpenVMS?
    
    Personally, I would say no. A group here at the center has a
    configuration running on a VAX 6620 that has a total of 198 nodes
    configured.  Their complaints with 1.6 have centered around the C3 but
    the controllers have been stable. Before 1.6 they had cronic problems
    with the line controllers that were caused by several know problems
    that have been fixed in 1.6. 
    
    >Are there any pans to support failover.
    
    You need to ask the product manager, Rae Kung - MUZICK::KUNG.
    
    Prior to the ECO, we had one problem that had similar
    symptoms such as the hung controller process reported in the base note. 
    This problem was only seen during and Archive though and is fixed in
    ECO1 for V1.6. As of that kit all the "hung controller" problems have
    been fixed so the base noter either has some environmental problem or
    has found something new. We really need to access the PCM engine when
    it breaks again.
    
    REgards,
       Dan Butterworth

    Any opinions on Alpha VMS vs. VAX VMS?
    
    The customer can replace the 4000/60 with a Sandpiper, would that be
    worth the effort?
    
    -cw

cw,

  The code base for Alpha and VAX is identical, so there shouldnt be any
  problems there, however an Alpha will give a significant performance boost 
  for your customer, I would advise them to upgrade if they are able to.

Cheers,
Phil