Get process information (the command, etc) from pid

Question

I have this code:

     for job in `jobs -p`; do
         code=0;
         wait ${job} || code=$?
         if [[ "${code}" != "0" ]]; then
            echo "At least one job failed with exit code => ${code}" ;
            exit 1;
         fi
     done

right after the line: echo "At least one job failed with exit code => ${CODE}" ;

I want to add a line that logs the command that failed, something like this:

     for job in `jobs -p`; do
         code=0;
         wait ${job} || code=$?
         if [[ "${code}" != "0" ]]; then
            echo "At least one job failed with exit code => ${code}" ;
            echo "The job that failed was $(ps -p ${job})"
            exit 1;
         fi
     done

not that a job is just the pid of the subprocess. The problem is that this line:

echo "The job that failed was $(ps -p ${job})"

doesn't really about anything - I need the actually command that was run, given the pid=job.

Answer 1

At the point you're running the ps command, the process already exited (you're handling its return code, right?)

Just change the logic so you run the ps command earlier in the script, before the wait and store that in a variable:

for job in $(jobs -p); do
    process_info=$(ps -p "${job}" -o args=)
    wait "${job}" || {
        code=$?
        echo "Job running [${process_info}] failed with exit code ${code}." >&2
        exit 1
    }
done

Answer 2

As pointed out, you can't use ps on a non-existing process.

If you have a bash that supports associative arrays:

declare -A jobinfo

# collect job information
joblist=( $(jobs -p) )
for pid in "${joblist[@]}"; do
    jobinfo["$pid"]="$( ps -p "$pid" -o args= )"
done

# later...

# wait for jobs
err=0
for pid in "${joblist[@]}"; do
    if ! { wait "$pid" || ! (( code=$? )); }; then
        printf 'Got non-zero exit status (%d) from PID %d: %s\n' \
            "$code" "$pid" "${jobinfo[$pid]}"
        err=1
    fi
done

# exit 0 if all was good, exit 1 otherwise
exit "$err"

Here I collect all the necessary data before starting to wait for jobs to exit. I also wait for all jobs to finish so that the user gets information about not just the first failing job.

---

Without using associative arrays (should work in the default bash on macOS):

# collect job information
joblist=( $(jobs -p) )
for pid in "${joblist[@]}"; do
    jobinfo+=( "$( ps -p "$pid" -o args= )" )
done

# later...

err=0
i=0
for pid in "${joblist[@]}"; do
    if ! { wait "$pid" || ! (( code=$? )); }; then
        printf 'Got non-zero exit status (%d) from PID %d: %s\n' \
            "$code" "$pid" "${jobinfo[$i]}"
        err=1
    fi

    i=$(( i + 1 ))
done

exit "$err"

Answer 3

If your OS (e.g. Linux) has a /proc directory, you can get every needed information from /proc/$PID

for PID in $(jobs -p); do
    dosomethingwith /proc/$PID/ # please take a look at what it contents ;-)
    done