Number of distinct users on linux/unix server

Today I was wondering if there was a way to figure out how many users were using a particular linux server. A simple thought that first came to my mind was couting the number of users in the output of 'ps -ef' command.

A simple 'wc -l' would give the number of lines in the output of 'ps -ef'.

The next step would be to just get the first column to get rid of the duplicate usernames.
'ps -ef | cut -d' ' f1' gave me the required output. Now the next step was to get only the unique entries. And the 'uniq' built in utility came to rescue.


The final command : 'ps -ef | cut -d' ' -f1 | sort | uniq | wc -l'

08/14/2010: I just realised that sort has an inbuilt option of unique
Instead of redirecting the out of sort to unique, one can use 'sort -u'

08/14/2010: Also instead of using 'ps -ef' command, one can use 'w'
Note: This command takes into account of the header row too. So the count would be count-- :)
08/14/2010: grep has an inbuilt option of '-v'which allows for invert matching. So the previous problem noted can be removed by adding 'grep -v UID'



The revised final command : 'w| cut -d' ' -f1 | sort -u | grep -v USER | wc -l'

08/14/2010: One final issue that I still see is the first line is taken into account here apart from the header.
 17:54:01 up 39 days, 10:20, 359 users,  load average: 0.59, 0.28, 0.15
USER     TTY      FROM              LOGIN@   IDLE   JCPU   PCPU WHAT