ps and top for Human Beings
ps and top for Human Beings
For a standalone install, copy / paste this command into a terminal:
curl -Ls https://github.com/walles/px/raw/python/install.sh | bash
Or you can install from Pypi:
sudo pip install --upgrade pxpx
Now, you should be able to run px, px --help or ptop from the command line. Otherwise please verify that /usr/local/bin is in your $PATH.
To try px without installing it, just download the latest px.pex, chmod a+x px.pex, then run ./px.pex.
Just type px or ptop, that’s a good start!
To exit ptop, press “q”.
Also try px --help to see what else px can do except for just listing all processes.
Running just px lists all running processes, with the most interesting ones last. Output truncated for brevity.
PID COMMAND USERNAME CPU RAM COMMANDLINE 0 kernel root -- -- kernel PID 0 116 KernelEventAgent root 0.01s 0% /usr/sbin/KernelEventAgent 426 corestoragehelperd root 0.05s 0% corestoragehelperd 808 PluginXPCService johan 0.05s 0% /System/Library/PrivateFrameworks/StoreXPCServices.framework/Versions/A/XPCServices/com.apple.appsto ... 80678 -fish johan 0.25s 0% -fish 80490 Google Chrome Helper johan 14.76s 1% /Applications/Google Chrome.app/Contents/Versions/58.0.3029.110/Google Chrome Helper.app/Contents/MacOS/Google Chrome Helper --type=renderer --field-trial-handle=1 --lang=sv --enable-offline-auto-reload --enable-offline-auto-reload-visible-only --blink-settings=disallowFetchForDocWrittenScriptsInMainFrame=false,disallowFetchForDocWrittenScriptsInMainFrameOnSlowConnections=false --enable-pinch --num-raster-threads=2 --enable-zero-copy --enable-gpu-memory-buffer-compositor-resources --enable-main-frame-before-activation --content-image-texture-target=0,0,3553;0,1,3553;0,2,3553;0,3,3553;0,4,3553;0,5,3553;0,6,3553;0,7,3553;0,8,3553;0,9,3553;0,10,34037;0,11,34037;0,12,34037;0,13,3553;0,14,3553;0,15,3553;1,0,3553;1,1,3553;1,2,3553;1,3,3553;1,4,3553;1,5,3553;1,6,3553;1,7,3553;1,8,3553;1,9,3553;1,10,34037;1,11,34037;1,12,34037;1,13,3553;1,14,3553;1,15,3553;2,0,3553;2,1,3553;2,2,3553;2,3,3553;2,4,3553;2,5,3553;2,6,3553;2,7,3553;2,8,3553;2,9,3553;2,10,34037;2,11,34037;2,12,34037;2,13,3553;2,14,3553;2,15,3553;3,0,3553;3,1,3553;3,2,3553;3,3,3553;3,4,3553;3,5,34037;3,6,3553;3,7,3553;3,8,3553;3,9,3553;3,10,3553;3,11,3553;3,12,34037;3,13,3553;3,14,34037;3,15,34037;4,0,3553;4,1,3553;4,2,3553;4,3,3553;4,4,3553;4,5,34037;4,6,3553;4,7,3553;4,8,3553;4,9,3553;4,10,3553;4,11,3553;4,12,34037;4,13,3553;4,14,34037;4,15,34037 --disable-accelerated-video-decode --disable-webrtc-hw-vp8-encoding --renderer-client-id=1309 80729 mdworker johan 0.06s 0% /System/Library/Frameworks/CoreServices.framework/Frameworks/Metadata.framework/Versions/A/Support/mdworker -s mdworker -c MDSImporterWorker -m com.apple.mdworker.shared 80728 AddressBookSourceSync johan 0.28s 0% /System/Library/Frameworks/AddressBook.framework/Versions/A/Helpers/AddressBookSourceSync.app/Contents/MacOS/AddressBookSourceSync 80720 GradleWrapperMain johan 2.6s 1% /usr/bin/java -Xdock:name=Gradle -Xdock:icon=/Users/johan/src/flickr-uploader/FlickrUploaderAndroid/media/gradle.icns -Dorg.gradle.appname=gradlew -classpath /Users/johan/src/flickr-uploader/FlickrUploaderAndroid/gradle/wrapper/gradle-wrapper.jar org.gradle.wrapper.GradleWrapperMain build 80727 GradleDaemon johan 25.56s 2% /Library/Java/JavaVirtualMachines/jdk1.8.0_60.jdk/Contents/Home/bin/java -Xmx1536M -Dfile.encoding=UTF-8 -Duser.country=SE -Duser.language=sv -Duser.variant -cp /Users/johan/.gradle/wrapper/dists/gradle-3.5-all/7s64ktr9gh78lhv83n6m1hq9u6/gradle-3.5/lib/gradle-launcher-3.5.jar org.gradle.launcher.daemon.bootstrap.GradleDaemon 3.5
- To give you the most interesting processes close to your next prompt, px
puts last in its output processes that:
- Have been started recently (can be seen in the list as high PIDs)
- Are using lots of memory
- Have used lots of CPU time
- Java processes are presented as their main class (GradleDaemon) rather than as their executable (java). This support is available for many VMs.
This lists all Java processes. Note how they are presented as their main class (GradleDaemon) rather than as their executable (java). This support is available for many VMs.
PID COMMAND USERNAME CPU RAM COMMANDLINE 80727 GradleDaemon johan 1m46s 5% /Library/Java/JavaVirtualMachines/jdk1.8.0_60.jdk/Contents/Home/bin/java -Xmx1536M -Dfile.encoding=UTF-8 -Duser.country=SE -Duser.language=sv -Duser.variant -cp /Users/johan/.gradle/wrapper/dists/gradle-3.5-all/7s64ktr9gh78lhv83n6m1hq9u6/gradle-3.5/lib/gradle-launcher-3.5.jar org.gradle.launcher.daemon.bootstrap.GradleDaemon 3.5 81192 GradleWrapperMain johan 2.57s 1% /usr/bin/java -Xdock:name=Gradle -Xdock:icon=/Users/johan/src/flickr-uploader/FlickrUploaderAndroid/media/gradle.icns -Dorg.gradle.appname=gradlew -classpath /Users/johan/src/flickr-uploader/FlickrUploaderAndroid/gradle/wrapper/gradle-wrapper.jar org.gradle.wrapper.GradleWrapperMain build
This lists all processes owned by the _coreaudiod user.
PID COMMAND USERNAME CPU RAM COMMANDLINE 204 DriverHelper _coreaudiod 6.67s 0% /System/Library/Frameworks/CoreAudio.framework/Versions/A/XPCServices/com.apple.audio.DriverHelper.xpc/Contents/MacOS/com.apple.audio.DriverHelper 195 coreaudiod _coreaudiod 43m10s 0% /usr/sbin/coreaudiod
sudo px 80727
This shows detailed info about PID 80727.
/Library/Java/JavaVirtualMachines/jdk1.8.0_60.jdk/Contents/Home/bin/java -Xmx1536M -Dfile.encoding=UTF-8 -Duser.country=SE -Duser.language=sv -Duser.variant -cp /Users/johan/.gradle/wrapper/dists/gradle-3.5-all/7s64ktr9gh78lhv83n6m1hq9u6/gradle-3.5/lib/gradle-launcher-3.5.jar org.gradle.launcher.daemon.bootstrap.GradleDaemon 3.5 kernel(0) root launchd(1) root --> GradleDaemon(80727) johan 31m33s ago GradleDaemon was started, at 2017-06-18T13:47:53+02:00. 7.6% has been its average CPU usage since then, or 2m22s/31m33s Other processes started close to GradleDaemon(80727): -fish(80678) was started 9.0s before GradleDaemon(80727) iTerm2(80676) was started 9.0s before GradleDaemon(80727) login(80677) was started 9.0s before GradleDaemon(80727) mdworker(80729) was started just after GradleDaemon(80727) mdworker(80776) was started 21.0s after GradleDaemon(80727) Users logged in when GradleDaemon(80727) started: _mbsetupuser johan 2017-06-18T14:19:26.521988: Now invoking lsof, this can take over a minute on a big system... 2017-06-18T14:19:27.070396: lsof done, proceeding. File descriptors: stdin : [PIPE] <not connected> (0x17d7619d3ae04819) stdout: [CHR] /dev/null stderr: [CHR] /dev/null Network connections: [IPv6] *:56789 (LISTEN) [IPv6] *:62498 (LISTEN) Inter Process Communication: mDNSResponder(201): [unix] ->0xe32cbd7be6021f1f For a list of all open files, do "sudo lsof -p 80727", or "sudo watch lsof -p 80727" for a live view.
- The command line has been split with one argument per line. This makes long command lines readable.
- The process tree shows how the Gradle Daemon relates to other processes.
- Details on how long ago Gradle Daemon was started, and how much CPU it has been using since.
- A list of other processes started around the same time as Gradle Daemon.
- A section describing where the standard file descriptors of the process go.
- A list of users logged in when the Gradle Daemon was started.
- A list of network connections the process has open.
- The IPC section shows that the Gradle Daemon is talking to mDNSResponder using Unix domain sockets.
The IPC data comes from lsof. sudo helps lsof get more detailed information; the command will work without it but might miss some information.
- Note how the default sort order of CPU-usage-since-ptop-started makes the display rather stable.
- Note the system load bar that correlates the system load with the number of CPU cores in the system. Green is load handled by physical cores, yellow (not shown here) is load handled by hyperthreading cores, and red is load over the number of cores.
- Note the fifteen minute load history graph in the load bar. On this system the load has been high for the last fifteen minutes. This is a visualization of the numbers you get from uptime.
- Note that after you press q to quit, the display is retained and some lines at the bottom are removed to prevent the information you want from scrolling out of view.
Note that the build infrastructure here is designed to create a px.pex file that works on both Python 2 and Python 3, and on “all” machines with Python interpreters (tested on Linux and OS X).
- Clone: git clone firstname.lastname@example.org:walles/px.git ; cd px
- Build and test: ./test.sh
- Run: ./px.pex
- To add dependencies, edit requirements.txt
- To run the same testing that CI does: ./ci.sh
Releasing a new Version
- Consider updating the Output section, push those changes.
- Do git tag and think about what the next version number should be.
- Do git tag --annotate 1.2.3 to set the next version number. The text you write for this tag will show up as the release description on Github, write something nice! And remember that the first line is the subject line for the release.
- git push --tags
- Go to the Releases page on GitHub, click your new release, click the Edit tag button, then attach your px.pex file that you just built to the release.
- Press “Publish Release”
Your release should now be visible on the pxpx page on Pypi.
- Store the output of lsof -F fnaptd0i from a big system in lsof.txt.
- ./tests/benchmark_ipcmap.py lsof.txt
Keeping this benchmark performant is important to be able to use px on big systems.
TODO top replacement
- Disable terminal line wrapping for smoother handling of terminal window resizes.
TODO iotop replacement
- When given the --top flag and enough permissions, record per process IO usage and present that in one or more columns.
- Details: When no users were found to be logged in at process start, automatically detect whether it’s because we don’t have history that far back or whether it seems to be that nobody was actually logged in. Inform the user about the outcome.
- In the px / top views, in the process owner column, maybe print other non-root process owners of parent processes inside parentheses?
- In the details report, if the current process has a working directory that isn’t /, list all other processes that have the same working directory.
- Ignore -E switch on Python command lines
- Make px list all processes with PID, owner, memory usage (in % of available RAM), used CPU time, full command line
- Output should be in table format just like top or ps.
- Output should be truncated at the rightmost column of the terminal window
- Output should be sorted by score, with score being (used CPU time) * (memory usage). The intention here is to put the most interesting processes on top.
- Each column should be wide enough to fit its widest value
- Add a section about installation instructions to this document.
- Add making-a-release instructions to this document
- Add a .travis.yml config to the project that: * OK: Runs flake8 on the code * OK: Tests the code on OS X * OK: Tests the code on Linux
- When piping to some other command, don’t truncate lines to terminal width
- If we get one command line argument, only show processes matching that string as either a user or the name of an executable.
- If we get something looking like a PID as a command line argument, show that PID process in a tree with all parents up to the top and all children down. This would replace pstree.
- If we get something looking like a PID as a command line argument, for that PID show: * A list of all open files, pipes and sockets * For each pipe / domain socket, print the process at the other end * For each socket, print where it’s going
- Doing px --version prints a git describe version string.
- Add a column with the name of each running process
- Put column headings at the top of each column
- In the details view, list processes as Name(PID) rather than PID:Name. To humans the name is more important than the PID, so it should be first.
- In the details view, list a number of processes that were created around the same time as the one we’re currently looking at.
- Implement support for px --top
- If the user launches px through a symlink that’s called something ending in top, enter top mode.
- top: On pressing “q” to exit, redraw the screen one last time with a few less rows than usual before exiting.
- top: Print system load before the process listing.
- Parse Java and Python command lines and print the name of the program being executed rather than the VM.
- In the details view, list users that were logged in when the process was started.
- In the details tree view, print process owners for each line
- Print $SUDO_USER value with process details, if set
- Run CI on both Python 2 and Python 3