Avahi/DBus: Shutdown of avahi stays undetected and leads to UnhandledPromiseRejection on shutdown.

[bauer-andreas]

Analysis

The current avahi advertiser implementation cannot detect if the avahi-deamon is rebooted or stopped. The service advertisement will be removed and will stay removed even if avahi starts again. Additionally, somewhere in the code is a UnhandledPromiseRejection triggered when HAP-NodeJS shuts down. Presumably this has to do with the advertiser trying to clean up its records, but failing to do so(?).

I haven't tested what happens if a txt record update is triggered while avahi is stoped or when avahi as restarted.

Expected Behavior

Ideally, we could detect a shutdown of avahi and also be notified once it is up and running again, such that we can readvertise our service records. I think, as soon as we detect that there is a running avahi instance, it is fine to assume it will be started at some point (e.g. might just be a restart).

Steps To Reproduce

1. Open mDNS monitoring app like Discovery
2. Start HAP-NodeJS with AVAHI advertiser
3. Observe service advertisement
4. Stop/restart avahi
5. See service advertisement being removed
6. ...
7. Stop HAP-NodeJS and observe the promise rejection warning below.

Logs

Received MESSAGE: {"serial":6,"destination":":1.190","errorName":"org.freedesktop.DBus.Error.UnknownObject","replySerial":6,"signature":"s","sender":":1.191","type":3,"flags":1,"body":["Method \"Free\" with signature \"\" on interface \"org.freedesktop.Avahi.EntryGroup\" doesn't exist\n"]}

(node:9242) UnhandledPromiseRejectionWarning: [object Array]
    at emitUnhandledRejectionWarning (internal/process/promises.js:170:15)
    at processPromiseRejections (internal/process/promises.js:247:11)
    at processTicksAndRejections (internal/process/task_queues.js:96:32)
(node:9242) UnhandledPromiseRejectionWarning: Unhandled promise rejection. This error originated either by throwing inside of an async function without a catch block, or by rejecting a promise which was not handled with .catch(). To terminate the node process on unhandled promise rejection, use the CLI flag `--unhandled-rejections=strict` (see https://nodejs.org/api/cli.html#cli_unhandled_rejections_mode). (rejection id: 1)
(node:9242) [DEP0018] DeprecationWarning: Unhandled promise rejections are deprecated. In the future, promise rejections that are not handled will terminate the Node.js process with a non-zero exit code.
    at emitDeprecationWarning (internal/process/promises.js:180:11)
    at processPromiseRejections (internal/process/promises.js:249:13)
    at processTicksAndRejections (internal/process/task_queues.js:96:32)

Configuration

-

Environment

• OS: Linux 5.10.63-v7+
• Software: 0.10.0-beta.7
• Node: v14.18.2
• npm: 6.14.15

Process Supervisor

not applicable

Additional Context

The Received MESSAGE ... log output is custom made, and prints anything arriving on the bus.connection.on("message") event.

[bauer-andreas]

CC: @adriancable, @oznu

[adriancable]

Hi @Supereg - I would like to re-open this and provide a suggested solution, please. I actually think this is quite important. If avahi crashes for any reason, then it will be restarted by systemd but HAP-NodeJS will not know, and so cannot re-advertise - so from the user's perspective the service will just disappear.

The solution is something like this in the constructor for AvahiAdvertiser:

this.bus.getService('org.freedesktop.Avahi').getInterface('/', 'org.freedesktop.Avahi.Server', (err, iface) => {
  if (iface) {
    iface.on('StateChanged', state => {
      if (this.path && state == 2) {
        // Already advertising, so need to re-advertise on Avahi restart
        this.startAdvertising();
      }
    });
  }
});

The states are:

0: collision 1: registering 2: running 3: connecting 4: failure

When Avahi starts up, it transitions to state 1 then state 2. So catching StateChanged to 2 will happen whenever Avahi is (re)started and is the correct signal to (re)advertise.

[adriancable]

@Supereg - thanks. Are you OK to add this code snippet to a beta and try it out? (I'm on the road for the next few days, so won't be able to make a PR ...)

[bauer-andreas]

@Supereg - thanks. Are you OK to add this code snippet to a beta and try it out? (I'm on the road for the next few days, so won't be able to make a PR ...)

0.10.0 beta is frozen as we prepare for the release. Don't really want to push last minute changes. But will include it in the oncoming beta cycle.

[adriancable]

@Supereg - perfect!

[github-actions[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[github-actions[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[bauer-andreas]

@adriancable

Where do you got the state id mapping from, looking at https://github.com/lathiat/avahi/blob/fd482a74625b8db8547b8cfca3ee3d3c6c721423/avahi-common/defs.h#L220-L227 I would have driven a different mapping.

Namely: 0: invalid 1: registering 2: running 3: collision 4: failure

See also https://github.com/lathiat/avahi/blob/fd482a74625b8db8547b8cfca3ee3d3c6c721423/avahi-daemon/dbus-protocol.c#L1211-L1238 where the StateChanged event is emitted.

[bauer-andreas]

Addressing issue with #970, review and verification pending.

[bauer-andreas]

This issue is now resolved in HAP-NodeJS version 0.11.0-beta.9.