perl one liners + replace complex string with key with any integer value

we are trying to remove completely the word - -XX:CMSInitiatingOccupancyFraction=50 from the file exampl.txt ( when value 50 could be any integer value )

for this task we used the following perl one liners , but without success

 perl -pi -e 's/\s*\K\Q-XX:CMSInitiatingOccupancyFraction=(^|\s)d{1,2}\E\s*//' exampl.txt
 perl -pi -e 's/\s*\K\Q-XX:CMSInitiatingOccupancyFraction=d{1,2}\E\s*//' exampl.txt 
 perl -pi -e 's/\s*\K\Q-XX:CMSInitiatingOccupancyFraction=[[:digit:]\E\s*//' exampl.txt 

example of the file:

SHARED_HADOOP_DATANODE="-server -XX:ParallelGCThreads=8 -XX:+UseConcMarkSweepGC -XX:CMSInitiatingOccupancyFraction=70 -XX:+UseCMSInitiatingOccupancyOnly -Xms{{namenode_heapsize}}" 

any suggestion what is wrong with my perl syntax?

perl -i -pe's/-XX:CMSInitiatingOccupancyFraction=\S+\s*//' exampl.txt

There are a number of problems in your attempts.

• \Q...\E escapes all non-word characters except those processed during the string-building phase ($, \E, \l, \L, \u, \U and the delimiter). You absolutely don't want this. \Q...\E is almost exclusively used as \Q$text\E.

• d{1,2} should be \d{1,2}. I switched to \S+ because there's no reason to be more specific.

• [[:digit:] should be [[:digit:]]+. \d is preferred over [:digit:], giving us [\d]+, which simplifies to \d+.

• ^ doesn't make sense there. It will never match since it matches the start of the string (without the m modifier) or the start of a line (with m).

• The leading \s*\K is useless.