Overview

Git commit signing fails because GPG cannot access its GPG key database due to a lock held by process ID <random-process-id-number> (in my case it’s ID 101853). It repeatedly shows “database_open waiting for lock” messages and eventually gets timeout on GPG key search and signing. In simple terms, this error simply blocks me from signing commits.

Why it happens?

GPG created a temporary lock files in ~/.gnupu and the extension of the files ends with *.lock*. I don’t know the exact reason, but if you are facing the same problem as me then it’s definitely because of this lock files and a background process is running behind which is keeping the files intact. As far as I have noticed, it seems like GPG somehow crashed and the process leaves the stale locks behind. Which is preventing us from signing commit using the gpg keys.

How to diagnose?

I can’t find the process ID 101853 to terminate, but there’s process running behind and I can’t find any GPG services running behind as well:

iamyaash@fedora:~/Projects/codeberg/forgejo-official$ ps -p 101853
   PID TTY          TIME CMD
iamyaash@fedora:~/Projects/codeberg/forgejo-official$ psgrep gpg
iamyaash   33108  0.0  0.0 231256  2328 pts/3    S+   17:24   0:00  |       \_ grep --color=auto gpg

So I tried looking into the files ends with *.lock in ~/.gnupg and I found a file named pubring.db.lock which is the culprit here:

iamyaash@fedora:~/Projects/codeberg/forgejo-official$ find ~/.gnupg/ -name '*.lock'
/home/iamyaash/.gnupg/public-keys.d/pubring.db.lock

I found out that deleting the file didn’t help either and the issue still persist.

iamyaash@fedora:~/Projects/codeberg/forgejo-official$ find ~/.gnupg/ -name '*.lock' -delete
/home/iamyaash/.gnupg/public-keys.d/pubring.db.lock

iamyaash@fedora:~/Projects/codeberg/forgejo-official$ echo "test" | gpg --clearsign
gpg: Note: database_open 134217901 waiting for lock (held by 101853) ...
gpg: Note: database_open 134217901 waiting for lock (held by 101853) ...
gpg: Note: database_open 134217901 waiting for lock (held by 101853) ...
gpg: Note: database_open 134217901 waiting for lock (held by 101853) ...
gpg: Note: database_open 134217901 waiting for lock (held by 101853) ...
gpg: keydb_search failed: Connection timed out
gpg: no default secret key: Connection timed out
gpg: [stdin]: clear-sign failed: Connection timed out

How to fix?

The primary reason for this issue is because, locks from crashed session kept the keys locked even after PID 101853 killed/terminated.

The typical solution seemed to be the full reset of ‘gpgconf’.

gpgconf --kill gpg-agent
gpgconf --kill keyboxd
pkill -f gpg-agent
pkill -f keyboxd

gpgconf --reload gpg-agent
gpgconf --reload scdaemon

This simple reset and reload seems to fix the issue:

iamyaash@laptop:~/Projects/gh/iamyaash/stashed$ echo "test" | gpg --clearsign
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

test
-----BEGIN PGP SIGNATURE-----
    ...
    ...
    ...
    ..
-----END PGP SIGNATURE-----