Is anyone running Progress in Solaris containers?

[nedthomas]

Is anyone running Progress in Solaris containers?

I'm trying to run Progress 8.3 databases in a Solaris 8 branded container using SUN M3000 servers (Solaris 10).

We want to rehost Progress legacy apps from SUN V490 SPARC servers (Solaris 8) to SUN M3000s running Solaris 10 using Solaris 8 containers. I am getting 1680 errors. We have 25 dbs (Progress 8.3E04 - 2003). I am not using port #s below 2000. I can start the db in single-user mode on the container. I've tried creating and running the small demo db but I get the same error in multiuser mode (container). I can start the db in multiuser mode in the global zone (Solaris 10). I've tried to use truss and view the strings from the core dump.

[FONT=r_ansi]root@usatxmdrx151/usr/local/db$ 18:34:58 BROKER 0: SYSTEM ERROR: Unable to kill[/FONT]
[FONT=r_ansi] parent process, errno= 4. (1680)[/FONT]
[FONT=r_ansi]18:34:58 BROKER 0: SYSTEM ERROR: The broker is exiting unexpectedly, beginning [/FONT]
[FONT=r_ansi]Abnormal Shutdown. (5292)[/FONT]
[FONT=r_ansi]18:34:58 BROKER 0: drexit: Initiating Abnormal Shutdown[/FONT]
[FONT=r_ansi]18:34:58 BROKER 0: ** Save file named core for analysis by Progress Software Co[/FONT]
[FONT=r_ansi]rporation. (439)[/FONT]
[FONT=r_ansi]18:34:58 BROKER 0: Begin ABNORMAL shutdown code 2 (2249)[/FONT]
[FONT=r_ansi]18:34:59 BROKER 0: Multi-user session end. (334)[/FONT]
[FONT=r_ansi] [/FONT]
##

[FONT=r_ansi][FONT=r_ansi]root@usatxmdrx151/usr/local/db$ more protrace.4549

PROGRESS stack trace as of Sun Sep 19 15:57:00 2010
Command line arguments are
/usr/dlc/bin/_mprosrv -pf /usr/local/db/test.pf
uttrace() +0xd0 from: /usr/dlc83d/bin/_mprosrv
utcore() +0xd0 from: /usr/dlc83d/bin/_mprosrv
drexit() +0x3ac from: /usr/dlc83d/bin/_mprosrv
msgout() +0x270 from: /usr/dlc83d/bin/_mprosrv
msgn() +0xac from: /usr/dlc83d/bin/_mprosrv
drkillparent() +0x60 from: /usr/dlc83d/bin/_mprosrv
doserve() +0x154 from: /usr/dlc83d/bin/_mprosrv
main() +0x88 from: /usr/dlc83d/bin/_mprosrv
_start() +0xdc from: /usr/dlc83d/bin/_mprosrv
??() +0xff3b5728 from: ??

##

[FONT=r_ansi][FONT=r_ansi]root@usatxmdrx151/usr/dlc$ more version
echo PROGRESS PATCH Version 8.3E04 as of July 22, 2003

[/FONT][/FONT][/FONT][/FONT]

 

[RealHeavyDude]

We've been using Solaris 10 zones for some 3 years now without any problems. The first release of Progress we were running that way was OpenEdge 10.0B. Progress 8 is more than 15 years old software and I suspect there is something in it which tries to access the hardware directly when a broker get's started. That could be something built into the executable itself or a parameter you've supplied to the server startup, like -directio.

HTH, RealHeavyDude.

 

[nedthomas]

The solution was found in this article - Solaris 8 brandz getppid hack.
http://onut.net/blog/archives/577

I was able to start our 25 Progress 8.3E04 databases with this hack. I need to thank Mick because it saved this rehost to containers project. I wanted to run Progress 8.3 under Solaris 10 given that we need to retest everything with the apps/dbs, but managers want to stay with Solaris 8 on new hardware. Thanks Ned