Diagnose and manage processes

Processes

ps: reports a snapshot of the current processes

ps # processes of which I'm owner

ps aux # all processes

It will print:

  • user, user owning the process
  • pid, process ID of the process (it is set when process start, this means that implicitly provides info on starting order of processes)
  • %cpu, the CPU time used divided by the time the process has been running
  • %mem, ratio of the process's resident set size to the physical memory on the machine
  • VSZ (virtual memory), virtual memory usage of entire process (in KiB)
  • RSS (resident memory), resident set size, the non-swapped physical memory that a task has used (in KiB)
  • tty, terminal the process is running on (? means that isn't attached to a tty)
  • stat, process state
  • start, starting time or date of the process
  • time, cumulative CPU time
  • command, command with all its arguments (those within [ ] are system processes or kernel thread)

Examples:

ps -eo pid,ppid,cmd,%cpu,%mem --sort=-%cpu

where:

  • -e, shows same result of -A
  • -o, specifies columns to show
  • --sort, sorts by provided parameter
ps -e -o pid,args --forest # the last arg shows a graphical view of processes tree

In /proc/[pid] there is a numerical subdirectory for each running process; the subdirectory is named by the process ID. The subdirectory /proc/[pid]/fd contains one entry for each file which the process has open, named by its file descriptor, and which is a symbolic link to the actual file. Thus, 0 is standard input, 1 standard output, 2 standard error, and so on.

Lists open files associated with process id of pid: lsof -p pid

Find a parent PID (PPID) from a child's process ID (PID): pstree -s -p <PID>

Background processes

Suffix command with & executes a process in background:

sleep 600 &
jobs    # lists processes in background
>[1]+  Running                 sleep 600 &
kill %1 # kills by job number
>[1]+  Terminated              sleep 600

To return a process in foreground: fg <PID>

Process priority

List "nice" value of processes: ps -e -o pid,nice,command Niceness (NI) value is a user-space concept, while priority (PR) is the process's actual priority that use by Linux kernel. In a Linux system priorities are 0 to 139 in which 0 to 99 for real time and 100 to 139 for users. Nice value range is -20 to +19 where -20 is highest, 0 default and +19 is lowest. A negative nice value means higher priority, whereas a positive nice value means lower priority.The exact relation between nice value and priority is:

PR = 20 + NI

so, the value of PR = 20 + (-20 to +19) is 0 to 39 that maps 100 to 139.

Note: Only root can assign negative values.

Execute a command in background with a given nice value to be added to the current one: nice -n <value> <command> &

Note: In case you want to associate a negative nice value to the process, then you'll have to use double hyphen: 

nice --10 wall <<end
System reboots in 5 minutes for Ubuntu Linux kernel update! 
Save all your work!!!
-- Sysadmin
end

Riassign priority to a process: renice -n <value> <pid>

Signals

Send a SIGTERM (15) signal to process: kill <pid>

Send a SIGKILL signal to process: kill -9 <pid>

Send a signal that correspond to number to process: kill -<number> <pid>

List all available signal and corresponding number: kill -l

Kill all child processes: pkill -P <ppid>

Kill all processes whose name matches a regex pattern: pkill -9 <pattern>

Kill by exact name (safer than pkill), unless -r is specified: killall <name>