[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference pamsrc::decmessageq

Title:	NAS Message Queuing Bus
Notice:	KITS/DOC, see 4.*; Entering QARs, see 9.1; Register in 10
Moderator:	PAMSRC::MARCUSEN

Created:	Wed Feb 27 1991
Last Modified:	Fri Jun 06 1997
Last Successful Update:	Fri Jun 06 1997
Number of topics:	2898
Total number of notes:	12363

2863.0. "dmqbcp startup problems + SIGINT woes" by SIOG::BATEMAN (We are all DECservers) Thu May 01 1997 13:08

    
    Periodically we are running into scenarios where the dmqbcp process
    won't startup as it claims there is already a dmqbcp running. Most of
    the time doing a rm -r /var/tmp/dmq clears the problem and dmqbcp starts 
    without trouble. On the very odd occasion this workaround around
    doesn't work as dmqbcp claims it is still running. Are there any other
    places to look for evidence of a now-dead dmqbcp process with V3.2? 
    
    On a related issue we have been talking about starting dmqgcp directly
    (ie: no dmqbcp process) but we are concerned about SIGINT signals
    killing the group if it is started interactively, eg: -
    
    # nohup dmqgcp ...  &
    # tail -f group.log
    Control-C			<=== Kills group (signal 2)
    
    Is there a better approach to starting the group interactively which
    prevents SIGINT killing the group? This happens on both V3.2 and
    V3.2A-ECO1.
    
    Thanks,
    Alan.

T.R	Title	User	Personal Name	Date	Lines
2863.1		XHOST::SJZ	Kick Butt In Your Face Messaging !	`Thu May 01 1997 18:24`	11
	what platform are you doing this on and what shell are you using ? if i start the dmqgcp process in the background (without nohup, just the &) and then I do a control-C from the same terminal that started the process, i don't run into any problems. the dmqgcp just keeps running. I know this to be true on all of the systems we have here. we use tsch (but I have gotten similar results using csh and sh). _sjz.
2863.2	Digital UNIX with sh or ksh	SIOG::BATEMAN	We are all DECservers	`Fri May 02 1997 04:48`	8
	Just tried with both sh and ksh on D UNIX and the SIGINT is detected by the gcp, ld, and journ processes. Also when the journ gets SIGINT it also seems to get a signal 15 (SIGTERM). BTW: I'm using /usr/bin/nohup is implemented as a shell command in some shell (csh for example). Alan.
2863.3		XHOST::SJZ	Kick Butt In Your Face Messaging !	`Fri May 02 1997 09:19`	6
	so start it from another terminal and log out, or don't press ctrl-c from that terminal. or start it from a script that uses csh. _sjz.
2863.4		SIOG::BATEMAN	We are all DECservers	`Fri May 02 1997 09:42`	2
	csh has the same behaviour. Maybe I'm better off using the dmqbcp for now.
2863.5		XHOST::SJZ	Kick Butt In Your Face Messaging !	`Fri May 02 1997 09:59`	2
	for V4.0 there is no bcp.
2863.6		SIOG::BATEMAN	We are all DECservers	`Fri May 02 1997 10:14`	11
	Is the real solution for dmqgcp to ignore SIGINT? I tried your suggestion of not using nohup but identical behaviour. I don't understand how using tcsh in your environment doesn't exhibit the same problem. Anyway, for now no ^C if you start dmqgcp. BTW: I understand that dmqbcp is retired in V4.0 but are there any reasons why it won't startup in a V3.2* system? Alan
2863.7		XHOST::SJZ	Kick Butt In Your Face Messaging !	`Fri May 02 1997 10:29`	12
	While you may not want the dmqgcp to go donw with SIGINT, other people do. Customers who really need that kind of protection for non-stop applications typically start their group as part of system startup. You might also try starting a group from the monitor. It may ex- hibit the same results; I don't know. But it is worth a try. If we do change it it won't get fixed before our next major re- lease. _sjz.