CDRIVER-3665 End-to-end test for OCSP caching

[bazile-clyde]

No description provided.

[kevinAlbs]

I wasn't sure how errexit treats background processes. To check, I created two files:

• fail.sh containing just exit 1
• run.sh containing:

set -o errexit
echo "Start"
./fail.sh &
echo "End (should fail sooner)" 

But run.sh prints both messages.

IIUC, calling wait on the background process PID may be all that's needed. That returns the exit code of that process. Modifying run.sh as follows seemed to give the desired behavior:

set -o errexit
echo "Start"
./fail.sh &
child_pid=$!
wait $child_pid
echo "End (should fail sooner)" 

With that, it terminated before printing the final message.

[bazile-clyde]

I'm not sure I understand. I think the first example is more in line with the expected behavior since this line doesn't terminate, i.e., return an exit code, in the background. It just hangs while the rest of the scripts continues. The sleep is just there to make sure that it can contact the OCSP responder before we kill the OCSP responder. But the program is still hanging around. Finally, kill will continue the program and return its exit code. If the test program does exit in the background for whatever reason this program still fails. Although I'm not exactly sure how I'd capture the exit status of a background that may either halt indefinitely or return.

[kevinAlbs]

Ah I was misunderstanding. I was thinking that if it terminated quickly, the script may quietly pass. But in that case the kill would fail

[kevinAlbs]

Ah I was misunderstanding. I was thinking that if it terminated quickly, the script may quietly pass. But in that case the kill would fail