Skip to content

Argonne Leadership Computing Facility

ATP and STAT on Theta


ATP and STAT are tools to debug abnormal program terminations such as segfaults. ATP (Abnormal Termination Processing) monitors a program while it runs. If the program crashes, ATP will invoke STAT (the Stack Trace Analysis Tool) to merge the stack backtraces of the application processes to an output file "". This merged stack backtrace file may then be visualized using STAT's visualization tool, stat-view.

Using ATP with stat-view

Scenario: When you try to run you get a segfault. After running, the job's `stderr file (which defaults to $COBALT_JOBID.error) contains:

user@thetalogin6:~> cat $COBALT_JOBID.error
_pmiu_daemon(SIGCHLD): [NID 03834] [c7-1c2s14n2] [Sat Aug 18 03:21:19 2018] \
                                   PE RANK 30 exit signal Segmentation fault
[NID 03834] 2018-08-18 03:21:19 Apid 4938801: initiated application termination
ATP and stat-view can be used to look into the segfault.

To use ATP, the ATP module should be loaded before linking your application . By default it is loaded on Theta, but to verify this, run module list, and check that the atp module is loaded.

user@thetalogin6:~> module list
Currently Loaded Modulefiles:
  1) modules/
  2) intel/ 
  3) craype-network-aries
  16) atp/2.1.2
  17) perftools-base/7.0.2

user@thetalogin6:~> make

Running the code

Next, the environment variable ATP_ENABLED must be set in the job script to enable ATP.

export ATP_ENABLED=1 # in bash
For Intel Fortran programs, also set the environment variable FOR_IGNORE_EXCEPTIONS in the job script:
export FOR_IGNORE_EXCEPTIONS=true # in bash
Then start your program with aprun as normal.
#COBALT -t 10 -n 1 -q debug-cache-quad

export ATP_ENABLED=1
aprun -n 64 -N 64 test.exe
If the program crashes, a should be produced, and the first process to die will write a stack backtrace to the job's stderr file (by default $COBALT_JOBID.error). For the scenario, this would look like:

user@thetalogin6:~> ls -ltr
-rw-r--r-- 1 user users      813 Aug 18 03:47
-rw-r--r-- 1 user users      768 Aug 18 03:48
-rw-r--r-- 1 user users     1742 Aug 18 03:48 262891.error
-rw-r--r-- 1 user users      143 Aug 18 03:49 262891.output
-rw-r--r-- 1 user cobalt    1965 Aug 18 03:49 262891.cobaltlog
user@thetalogin6:~> cat 262891.error
Application 4938811 is crashing. ATP analysis proceeding...

ATP Stack walkback for Rank 31 starting:
ATP Stack walkback for Rank 31 done
Process died with signal 11: 'Segmentation fault'
View application merged backtrace tree with: stat-view
You may need to: module load stat

_pmiu_daemon(SIGCHLD): [NID 03834] [c7-1c2s14n2] [Sat Aug 18 03:48:09 2018] \
                                               PE RANK 1 exit signal Killed
[NID 03834] 2018-08-18 03:48:09 Apid 4938811: initiated application termination

Looking at the output

The program stat-view (the Stack Trace Analysis Tool viewer) can be used to view the backtrace file, which shows what the program was doing at the time of abnormal termination.

user@thetalogin6:~> module load stat

user@thetalogin6:~> stat-view

The difference between and is that should contain line numbers.

The output will look like the following:

dot difference

Difference between and is that

This shows that there is a segfault at line 19 of test.c, the test file. Each box is typically a function in a stack backtrace or a comment, denoted by brackets ( "[ ]" ) and discussed later.

Moving from top to bottom, following the arrows, follows a backtrace. The arrow connecting the boxes is labeled with how many ranks are active for the function, and which ranks. Here we see that for every function, 64 ranks were active, and they are ranks 0-63 (this makes sense since this was done with a single node, 64 rank job). If only a subset of ranks had had a segault, then the graph would label how many segfaulted and which ones.

Some of the boxes are not functions, but are comments, which is denoted by the use of bracket ("[ ]") in the name. Here, the "[Fault Summary]" box is a summary tree saying that all 64 ranks, 0-63 faulted due to a SIGSEV.

Note: To use stat-view, an X display will need to be forwarded from your local machine. When you connect to the Theta login node, use the -Y or -X option to ssh:

user@thetalogin6:~> ssh -Y