4 ms·

On Linux I very regularly use 'strace', to see what a program is currently doing (in terms of syscalls), what files were opened/read/written/closed, etc.

thomasknowles 15y ago
strace is the Daddy of debugging especially when you can attach it to an already existing program using it's PID. I have solved so many permissions errors problems with strace where the application itself just failed without logging anything.
- zobzu 15y ago
  try austrace. also gdb of course.
- nikcub 15y ago
  and on OS X you have DTrace (from Solaris), which can be called using dtruss, which is a shell script wrapper around DTrace also, in developer > applications there is an app called 'Instruments' which is an excellent GUI front-end to DTrace
- Meai 15y ago
  If the application fails, wouldn't it just shut down and the process ends? Meaning that I can't call strace anymore? Do you have suggestions for that usecase? How would I know what the process id is going to be before I start the program?
  - tedunangst 15y ago
    $ strace program args
- nz 15y ago
  Similar experience here, except that I've solved similar issues (and far more complicated issues) with DTrace instead of strace. The range of information that DTrace can retrieve about a running application surpasses anything that strace can retrieve. DTrace can retrieve information on a systemic scale, whereas each strace instance operates on a single process. And the overhead is significantly lower with DTrace. For many of the systems applications I design for Illumos, I've used DTrace probes as a way of logging very frequent events, on demand (to avoid frequent IO). All of the events that _must_ be logged for the application to function properly, are logged and fsync'd. I think that in most systems dynamic tracing will eventually replace a significant portion of the logging functionality that people code into their applications. Either way, if you think strace is sweet, give DTrace a spin. Some helpful links: [0] A video demostration of DTrace, by the creator of DTrace. [0] http://www.youtube.com/watch?v=6chLw2aodYQ http://www.youtube.com/watch?v=6chLw2aodYQ [1] A post I wrote, demonstrating DTrace. Similar posts can be found on the blog's dtrace-addict page. [1] http://nickziv.wordpress.com/2011/04/08/adventures-of-a-dtrace-addict-part-0/ http://nickziv.wordpress.com/2011/04/08/adventures-of-a-dtra... [2] A wiki that contains DTrace examples for various languages and system facilities. Some examples may be Illumos-centric. [2] http://www.solarisinternals.com/wiki/index.php/DTrace_Topics http://www.solarisinternals.com/wiki/index.php/DTrace_Topics [3] DTrace's home. Contains blogs by the engineers behind DTrace. [3] http://www.dtrace.org http://www.dtrace.org UPDATE: Meant to reply to parent's parent.
  - tangothedog 15y ago
    systemtap is another utility in this vein that's worth checking out http://sourceware.org/systemtap/ http://sourceware.org/systemtap/
    - nz 15y ago
      I already checked it out and the timer probes don't work at all. On one system, I couldn't even invoke kernel functions. On another system, the kernel panicked as soon as I executed `stap -e ...`. Basically unusable, at this point. I hope it improves because developers on linux could really benefit from dynamic tracing.
falcolas 15y ago
strace, combined with lsof and netstat can help you with so many things it's not even funny. I've debugged stalled processes across 3 separate servers with these three tools alone.

thomasknowles 15y ago

strace is the Daddy of debugging especially when you can attach it to an already existing program using it's PID. I have solved so many permissions errors problems with strace where the application itself just failed without logging anything.

zobzu 15y ago
try austrace. also gdb of course.
nikcub 15y ago
and on OS X you have DTrace (from Solaris), which can be called using dtruss, which is a shell script wrapper around DTrace also, in developer > applications there is an app called 'Instruments' which is an excellent GUI front-end to DTrace
Meai 15y ago
If the application fails, wouldn't it just shut down and the process ends? Meaning that I can't call strace anymore? Do you have suggestions for that usecase? How would I know what the process id is going to be before I start the program?
- tedunangst 15y ago
  $ strace program args
nz 15y ago
Similar experience here, except that I've solved similar issues (and far more complicated issues) with DTrace instead of strace. The range of information that DTrace can retrieve about a running application surpasses anything that strace can retrieve. DTrace can retrieve information on a systemic scale, whereas each strace instance operates on a single process. And the overhead is significantly lower with DTrace. For many of the systems applications I design for Illumos, I've used DTrace probes as a way of logging very frequent events, on demand (to avoid frequent IO). All of the events that _must_ be logged for the application to function properly, are logged and fsync'd. I think that in most systems dynamic tracing will eventually replace a significant portion of the logging functionality that people code into their applications. Either way, if you think strace is sweet, give DTrace a spin. Some helpful links: [0] A video demostration of DTrace, by the creator of DTrace. [0] http://www.youtube.com/watch?v=6chLw2aodYQ http://www.youtube.com/watch?v=6chLw2aodYQ [1] A post I wrote, demonstrating DTrace. Similar posts can be found on the blog's dtrace-addict page. [1] http://nickziv.wordpress.com/2011/04/08/adventures-of-a-dtrace-addict-part-0/ http://nickziv.wordpress.com/2011/04/08/adventures-of-a-dtra... [2] A wiki that contains DTrace examples for various languages and system facilities. Some examples may be Illumos-centric. [2] http://www.solarisinternals.com/wiki/index.php/DTrace_Topics http://www.solarisinternals.com/wiki/index.php/DTrace_Topics [3] DTrace's home. Contains blogs by the engineers behind DTrace. [3] http://www.dtrace.org http://www.dtrace.org UPDATE: Meant to reply to parent's parent.
- tangothedog 15y ago
  systemtap is another utility in this vein that's worth checking out http://sourceware.org/systemtap/ http://sourceware.org/systemtap/
  - nz 15y ago
    I already checked it out and the timer probes don't work at all. On one system, I couldn't even invoke kernel functions. On another system, the kernel panicked as soon as I executed `stap -e ...`. Basically unusable, at this point. I hope it improves because developers on linux could really benefit from dynamic tracing.

falcolas 15y ago

strace, combined with lsof and netstat can help you with so many things it's not even funny. I've debugged stalled processes across 3 separate servers with these three tools alone.