Monitor sssd and its children

[Sebastian Stark]

Hi,

I wonder how to monitor sub processes like in the example given below:

# monit procmatch sssd
List of processes matching pattern "sssd":
------------------------------------------
        /usr/sbin/sssd
        /usr/lib/sssd/sssd/sssd_be --domain XXX
        /usr/lib/sssd/sssd/sssd_nss
        /usr/lib/sssd/sssd/sssd_pam
------------------------------------------
Total matches: 4

# pstree -A -p 1030
sssd(1030)-+-sssd_be(1032)
           |-sssd_nss(1044)
           `-sssd_pam(1045)

Sometimes a bug makes sssd_be crash with a segfault and sssd is not
able to restart it (it usually does). Now when that happens I want
"stop sssd; start sssd" to be executed. My humble way of doing this so
far is:

check process sssd matching "^sssd"
    start program = "/usr/local/sbin/restart-sssd.sh"
    stop program = "/sbin/stop sssd"
    if children < 3 then restart
    if 5 restarts within 5 cycles then timeout

which works, but I would like to know wether somebody has a better way
for this case. I would also rather not have to use the restart-sssd.sh
script for issueing stop/start.


Sebastian