bug report in command ( ps -ef | grep "myTask" | grep -v "grep" | wc -l )

[OSK Lu Jun]

To whom it may concern

I am developing a server application on Linux AS2.1(Japanese Version) with
Oracle9iR2.
Now I meet a very strange problem that I can not solve. I think maybe it is
a bug of Linux.
So I make a report to someone who is Linux professional here.

The details of the problem as follows,
=========================================================================
OS: Redhat Linux Advanced Server release 2.1AS(Pensacola)  2.4.9-e.3smp
Japanese Version
CPU: Intel Xeon 3.06GHz (1 CPU System)
RAM: 1.5G
Compiler: g++, Oracle ProC

Problem description:
There are 7 executable programs and 1 process-check bash program in my
server
application. When the application start up, these executable programs will
be loaded
and stationed in the RAM. I  use the process-check bash program to check if
they
are alive. In bash file I use the COMMAND
(  PROCESS_NUM=`ps -ef | grep  "myTask" | grep -v "grep" | wc -l` )
to judge the alive  status of my executable programs.

chkprocess() {
PROCESS_NUM=`ps -ef | grep  "myTask" | grep -v "grep" | wc -l`
   if [ $PROCESS_NUM -eq 0 ]; then
      return $NON_EXIST
   else
     return $EXIST
   fi
}

But I get a stange result in my system. Sometimes the return value of
PROCESS_NUM is 0, sometimes the return value of PROCESS_NUM is
not 0. After it happened, I check my program in B-shell  by 'ps x', the
result show that my programs are still alive, and then I  check my program's
log file. It also show my task programs have not been dead, still alive
healthily.
There are no restart functions in my programs.  I have let my programs
continue running 24 hours, this problem will happen 1 or 2 times in a day.

So I think the COMMAND in my bash file may have a bug. And  I wonder if
there
are other ways to judge a specified program's alive status. I am looking
forward
your replys.

Sincerely yours,
Lu,Jun