Upgrading syncthing-relaysrv to 1.15.0 fails in post-installation

[bronikowski]

• Versions: Upgrading the to the latest release (from candidate to stable) on ARM
• What happened: The dist-upgrade fails with an error in a post-installation script
• What was expected: your run-of-the-mill upgrade

Output:

Setting up syncthing-relaysrv (1.15.0) ...
/var/lib/dpkg/info/syncthing-relaysrv.postinst: 11: /var/lib/dpkg/info/syncthing-relaysrv.postinst: Syntax error: "else" unexpected
dpkg: error processing package syncthing-relaysrv (--configure):
 installed syncthing-relaysrv package post-installation script subprocess returned error exit status 2
Errors were encountered while processing:
 syncthing-relaysrv
E: Sub-process /usr/bin/dpkg returned an error code (1)

[calmh]

Far as I can tell we don't actually include a postinst script for the relay server. What does that script look like?

[bronikowski]

Sure, here's what I get when I less the file.

#!/bin/sh


after_upgrade() {
    :

systemctl --system daemon-reload >/dev/null || true
debsystemctl=$(command -v deb-systemd-invoke || echo systemctl)
if ! systemctl is-enabled strelaysrv >/dev/null
then
else
    $debsystemctl restart strelaysrv >/dev/null || true
fi
}

after_install() {
    :

systemctl --system daemon-reload >/dev/null || true
debsystemctl=$(command -v deb-systemd-invoke || echo systemctl)
}

if [ "${1}" = "configure" -a -z "${2}" ] || \
   [ "${1}" = "abort-remove" ]
then
    # "after install" here
    # "abort-remove" happens when the pre-removal script failed.
    #   In that case, this script, which should be idemptoent, is run
    #   to ensure a clean roll-back of the removal.
    after_install
elif [ "${1}" = "configure" -a -n "${2}" ]
then
    upgradeFromVersion="${2}"
    # "after upgrade" here
    # NOTE: This slot is also used when deb packages are removed,
    # but their config files aren't, but a newer version of the
    # package is installed later, called "Config-Files" state.
    # basically, that still looks a _lot_ like an upgrade to me.
    after_upgrade "${2}"
elif echo "${1}" | grep -E -q "(abort|fail)"
then
    echo "Failed to install before the post-installation script was run." >&2
    exit 1
fi

EDIT: it might be important, I have a running relay, so that if-block had to evaluate.

[calmh]

Apparently this is something auto generated by the packaging script because we say we have a systemd service. 🤷

[bronikowski]

Ah, the wonders of packaging. Clearly the if is there to prod the running process to HUP.

[imsodin]

jordansissel/fpm#1757

And the simple fix is pending for quite some time: jordansissel/fpm#1764

I'll use our own post-inst script like we already do for syncthing to fix this. Looks like relying on fpm features as little as possible is the way to go.

[js-d-coder]

I fixed this installation issue by adding echo after line 10 in /var/lib/dpkg/info/syncthing-relaysrv.postinst and then ran sudo apt-get update && sudo apt-get upgrade && sudo apt-get dist-upgrade && sudo apt-get install -f. Voila, it worked. ✌🏻