mjhjmtu

I have a process in a Linux installation that at some point has some kind of spike and passes the max allowed number of threads/processes allowed by the system. I found this by checking ps -elfT | wc -l repeatedly.

But what I don't know is what exactly is it that causes this spike.

The output of ps -elfT has a lot of information, but I cannot easily understand if there is some child process that does some kind of "blurp" in forking and makes a mess.

How could I figure that out?

Example: ps -elfT | cut -d' ' -f3 | sort |uniq gives me the processes running at the time. How could I add a count to see how much each contributes to the total?

ps -eo nlwp,pid,args --sort nlwp

Would show a list of processes sorted by their number of threads.

For a top-like view of that, you can always do:

watch -n 1 'ps -eo nlwp,pid,args --sort -nlwp | head'

Or you could use... top.

• press f to select the fields to display.


• locate nTH (the number of threads) and press d to display it and s to make it the sort order


• you can adjust its display position with → and then ↑ and ↓ and ⏎.


• 
  q to get back to the process list


• press H if you want to see all the threads.


• 
  d to adjust the delay.


• 
  ? for help.