'step ca renew' shell exit codes

[kotfu]

I use bacula to backup all the virtual machines in my home lab. I've had TLS running with all the bacula services for years, using a CA created by hand using openssl. I'm now migrating to step ca so I can more easily automate renewals. My bacula config uses 3 different certificates, with various combinations of these certificates used by 3 different bacula daemons.

I'm using a cron job to run a script to renew my certificates. The script uses step ca renew to attempt renewal of each of the 3 certificates. I need logic in my script to decide which daemons to restart depending on which of the certificates have been renewed. Because of the particulars of this use case, I would like to avoid using --exec so I don't have to restart these daemons multiple times.

With all that context, here's my question/idea. I want step ca renew to return a shell exit code of 0 if the certificate is renewed, a shell exit code of 1 if it isn't, and a shell exit code of 2 of there is a command line argument parsing problem. Right now, step ca renew returns an exit code of 0 whether the cert was renewed or not. As a result, I can only think of two ways to tell whether step ca renew has successfully renewed the certificate:

1. Hash the cert file before I run step ca renew, and hash it again after, if the hash isn't the same, then the cert has been renewed
2. Capture stderr and parse the message generated by step ca renew to figure out if it thinks it has written a new certificate

Both of them are possible, but a little janky. If step ca renew returned a different exit code on renewal or non-renewal, then you could write something like:

$ step ca renew --force file.crt file.key && logger -t step-ca renewing file.crt

which is really convenient. It also let's me do this in a script:

step ca renew --force file.crt file.key 2>/dev/null
if [ "$?" -eq 0 ]; then
    # do my fancy logic to figure out if this or other certs have been renewed and restart appropriate daemons
fi

[tashian]

Hi! Thanks for the suggestion, I've created issue #2158 from your question.