SUNW.gds resource creation problem / service startup failure

Question

We have a vendor provided set of application server start and stop scripts that we are trying to mold into a GDS resource. The scripts must be run as a user not as root. This app takes about 6 minutes to start up and it creates about 400 processes in doing so. We gave the user cluster.modify auth so it can create the resource, and the command takes but when the resource is being created, the startup script does not seem to fire. So I'm a little stumped as to make this thing run under the user account project settings. The other thing I noticed is the ports that I specify in the list are never present until the end of the 6 minutes of startup junk, and the GDS seems to log problems about these ports at the onset of creation. Here is the command for the create:
clrs create -g ss_wfm-rg -t SUNW.gds  \
-p Start_Command="/home/appl/advxrt/R8.0g/TOP/SOLARIS/bin/StartServer" \
-p Stop_Command="/home/appl/advxrt/R8.0g/TOP/SOLARIS/bin/ShutdownServer" \
-p Resource_dependencies=advxprd,ss_hasp-rs \
-p Port_list=22100/tcp,23100/tcp,27100/tcp,4100/tcp,17100/tcp \
-p Start_timeout=420 \
-p Stop_timeout=120 \
-p Child_mon_level=-1 \
ss_server_gds-rsAnd from the messages log...
Aug 18 07:31:04 prddsm01nd01 SC[SUNW.gds:6,ss_wfm-rg,ss_server_gds-rs,gds_svc_start]: [ID 141062 daemon.error] Failed to connect to host advxprd and port 221
00: Connection refused.
Aug 18 07:31:04 prddsm01nd01 SC[SUNW.gds:6,ss_wfm-rg,ss_server_gds-rs,gds_svc_start]: [ID 805735 daemon.error] Failed to connect to the host <advxprd> and po
rt <22100>.
Aug 18 07:31:06 prddsm01nd01 Cluster.PMF.pmfd: [ID 887656 daemon.notice] Process: tag="ss_wfm-rg,ss_server_gds-rs,0.svc", cmd="/bin/sh -c /home/appl/advxrt/R
8.0g/TOP/SOLARIS/bin/StartServer", Failed to stay up.
Aug 18 07:31:06 prddsm01nd01 Cluster.PMF.pmfd: [ID 534408 daemon.notice] "ss_wfm-rg,ss_server_gds-rs,0.svc" restarting too often ... sleeping 30 seconds.
Aug 18 07:31:06 prddsm01nd01 Cluster.RGM.rgmd: [ID 650825 daemon.error] Method <gds_svc_start> on resource <ss_server_gds-rs> terminated due to receipt of si
gnal <9>Basically I would like to know if the ports list should include all ports that the app will eventually listen on, or some, or minimal? The other biggy is how to get this thing to launch as a user and if it fail over to another node does not start as root. I tried doing an su - acctname -c "script" but it won't start it. Thanks!

Answer 1

Detlef or anyone familiar with GDS...
We agree here that customizing the GDS using your template is the way to go. What I have noticed is that since creating the wrapper scripts to call the newtask command, setup the environment, etc, when the wrapper script finishes (kicking off the native script) the RGM sees this as a fault and immeidately tries to restart the agent. Here is a snippet of the messages log.
Sep 24 13:48:03 devdsm01nd01 Cluster.PMF.pmfd: [ID 887656 daemon.notice] Process: tag="ss_stage-rg,ss_stage_server_gds-rs,0.svc", cmd="/bin/sh -c /home/appl/advxtst/tds/SS_kicker.sh", Failed to stay up.
Sep 24 13:48:03 devdsm01nd01 Cluster.RGM.rgmd: [ID 784560 daemon.notice] resource ss_stage_server_gds-rs status on node devdsm01nd01 change to R_FM_FAULTED
Sep 24 13:48:03 devdsm01nd01 Cluster.RGM.rgmd: [ID 922363 daemon.notice] resource ss_stage_server_gds-rs status msg on node devdsm01nd01 change to <Service daemon not running.>
Sep 24 13:48:03 devdsm01nd01 SC[SUNW.gds:6,ss_stage-rg,ss_stage_server_gds-rs,gds_probe]: [ID 670283 daemon.notice] Issuing a resource restart request because the application exited.So we figured out that if we suspend the RG, then we can of course stop and start the app all day long using the clrs dis|enable commands but is using these wrapper scripts still the way to go even with the GDS template? Even assigning the user's project settings to the resources does not seem to allow the native vendor supplied startup scripts to launch (they say it is purely because we did not setup the working directory, etc, correctly, which the wrapper scripts do quite well). Is is normal for the wrapper script to "die" but yet have GDS probe the ports after X amount of time? Is this where the sleeping dummy comes into play?
thanks in advance!