when application failed, it will not restart manually

[smkhairnar]

we are using service guard cluster. when package is failed , we tried to start manually by cmrunpkg on node . But it will unalble to restart package . after that i tried to halt the node but it also unable to halt, then tried to halt cluster still its unable to halt the cluster. why please provide solution solve to problem .

procedure are below which followed

cmrunpkg -pkgname ( unable to restart check syslog

cmhaltnode --- unable to restart check syslog

cmhaltcl -f -----unable to restart check syslog

pls provide solution to restart application without restart node , or restart cluster , or
restart server.


Thanks & regards

[Basheer_2]

Hello,

what are the logs saying

/var/adm/syslog/syslog.log
/etc/cmcluster/cluster-name/cmctl.log ( yoour path depends)

what does cmviewcl -v say

[stephen peng]

what do you mean package fail? post your syslog.log and (package_name)cntl.log and cmviewcl -v

[vasanth_2]

Hi smkhairnar,

Which version of the SG are you using and what is the version of OS?

Thanks
Vasanth

[Deeos]

hi can you do the following step
?

#cd /etc/cmcluster/package-name folder
#cmcheckconf -P packagename.conf


check consistancy of package configuration.
is your packagename.cntl file has recently modified?


#ll packagename.cntl


or you have option to take backup of packagename.cntl file and re-create packagename.cntl file and edit it as per your requiremants

#cmmakepkg -s packagename.cntl



because if package is niether halt nor start is causes by Corrupted packagename.cntl file



Regards
Deeos

[Stephen Doud]

smkhairnar,
As Basheer stated, Serviceguard puts information in /var/adm/syslog/syslog.log and /etc/cmcluster// or /var/adm/cmcluster/log/ for admins to inspect and learn why packages or the cluster are not working as expected. Most novices do not know where to look. Now that you know, you have access to the same information that experienced Serviceguard admins look at to learn what transpired.

My advice: look for the first error that occurs at, or just before the problem occurs. This error may result in a cascade of other errors that might mislead you, so work on the first error first. After solving it, the remaining errors may disappear after further testing.