[Klug-general] Checking For Duplicate Instance Of Script

[David Halliday]

*I have solved the strange increment problem!*

To demonstrate the issue I created this test script:
#!/bin/bash
#declare an int
declare -i INSTANCECOUNT

echo "basename: $(basename $0)"

echo "raw ps output command hard coded:"
ps aux | grep test3 | grep -v "grep" | wc -l

echo "raw ps output command basename:"
ps aux | grep $(basename $0) | grep -v "grep" | wc -l

echo "Command substitution test"
echo $(ps aux | grep $(basename $0) | grep -v "grep" | wc -l)

#assign & output 1:
INSTANCECOUNT=$(ps aux | grep $(basename $0) | grep -v "grep" | wc -l)
echo "Count 1: $INSTANCECOUNT"

#assign & output 2:
INSTANCECOUNT=$(ps aux | grep $(basename $0) | grep -v "grep" | wc -l
2>/dev/null)
echo "Count 2: $INSTANCECOUNT"

#assign & output 3:
INSTANCECOUNT=$(ps aux | grep test3 | grep -v "grep" | wc -l)
echo "Count 3: $INSTANCECOUNT"

#assign & output 4:
INSTANCECOUNT=$(ps aux | grep $(basename $0) | grep -v "grep" | wc -l | sed
-e "s/[^0-9]//g")
echo "Count 4: $INSTANCECOUNT"

This produced the output:
raw output command hardcoded:
1
raw output command:
1
Command substitution test
2
Count 1: 2
Count 2: 2
Count 3: 2
Count 4: 2

When I used another process, say grep for sql rather than $(basename $0)
the counts were as expected!
Then I twigged, this was an issue of the ps command running using command
substitution. In sohort by using command substitution "$()" a sub process
was being created this was in turn creating an extra entry within ps which
went as soon as the command substitution finished. Having done work with
sub processes and simultanious processing under bash before I should have
known. For those interested some links:

   - Subshells (slightly different but same behaviour in ps demonstrated):
   http://tldp.org/LDP/abs/html/subshells.html
   - bash internal variables ($BASHPID and $$ useful for looking at
   processes and sub processes by ID):
   http://tldp.org/LDP/abs/html/internalvariables.html

Note, $BASHPID is fairly new, only in bash v4 and up, to test:

echo $BASH_VERSINFO[0]
If you have over version 4 then you can test with this in a script:
#!/bin/bash
echo "Bash ID: $$ : $BASHPID : $BASH_SUBSHELL"
echo $(echo "Bash ID: $$ : $BASHPID : $BASH_SUBSHELL")

Interesting results, and worth thinking about if you are writing clever
scripts!

So, as a result I'm not 100% trusting of my script and now I understand a
little more why!
*
So, it's still on with the script/library!*
For the sake of sanity I have made the change to my test case.

controlLib.sh
=========
checkForDuplicateScript () {
  SCRIPTNAME=$(basename $0)
  #declare INSTANCECOUNT as
integer
  declare -i INSTANCECOUNT
  INSTANCECOUNT=$(ps aux | grep "$SCRIPTNAME" | grep -v "grep" | wc -l
2>/dev/null)
  if [ "$INSTANCECOUNT" -gt "2" ]
  then
    echo "Extra instance found of script: $SCRIPTNAME : Full ps output: "
    ps aux | grep "$SCRIPTNAME"
    echo "Instance Count: $INSTANCECOUNT : $BASHPID : $SCRIPTNAME"
    exit 1
  fi
}

Still produces the same behaviour, so I will presume that bash kindly
allows you to have overkill on function declaration (that's how I have been
getting away with it all this time, but being lazy will prevent that from
now on).

So, the game is still on!


On 21 December 2011 12:37, David Halliday <david.halliday at gmail.com> wrote:

> I'll give that a try but since all my other functions have been working as
> expected to date with that format I'll put it down as overkill.
> Functions Doc: http://tldp.org/LDP/abs/html/functions.html
>
> I'll give that a try to see if it's what is causing the strange behaviour.
>
>
> On 21 December 2011 12:27, Laurence Southon <laurence at southon.uk.net>wrote:
>
>> On 20/12/11 21:32, David Halliday wrote:
>> > function checkForDuplicateScript(){
>>
>> I don't know if it's the reason but that function declaration should be
>> either:
>>
>> function checkForDuplicateScript { ... }
>>
>> or
>>
>> checkForDuplicateScript () { ... }
>>
>> Not the keyword 'function' and '()' at the same time.
>>
>> LS
>> --
>> Laurence Southon
>> Tiger Computing, Bexley
>> www.tiger-computing.co.uk
>>
>> _______________________________________________
>> Kent mailing list
>> Kent at mailman.lug.org.uk
>> https://mailman.lug.org.uk/mailman/listinfo/kent
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.lug.org.uk/pipermail/kent/attachments/20111221/338ce398/attachment-0001.htm>