Add method to detect if network has changed and reinitialize zeroconf

[jamalex]

Summary

We identified that zeroconf doesn't pick up when a new network connection is established, so it doesn't become discoverable on networks that it wasn't connected to at the time Kolibri started up.

This PR adds a method for detecting when the set of bound IP addresses on the system has changed, and reinitializing the zeroconf internals to ensure they're attached to each of those IPs.

Note: no code has been added to actually call this new method. We may want to spawn a thread that calls it every minute (as it should be very lightweight most of the time, since it's checking the list of bound IPs against its cached copy and only doing something if it changes). We'll need to ensure it happens in the same process, as the zeroconf state is process-local (in global vars inside kolibri.core.discovery.utils.network.search), so we may not want to use iceqube.

Reviewer guidance

To test:

• Have Kolibri running on another machine on the network
• On the test machine, disconnect from the network (e.g. disable wifi)
• Start kolibri shell, and run from kolibri.core.discovery.utils.network.search import *; register_zeroconf_service(999)
• Run [p["device_name"] for p in get_peer_instances()], and you should only see your own machine name
• Reconnect to the network
• Run [p["device_name"] for p in get_peer_instances()] again several times, waiting in between, and you should still only see your own machine name
• Run reinitialize_zeroconf_if_network_has_changed()
• Start trying [p["device_name"] for p in get_peer_instances()] again. After a short time, it should start showing the name of the other device on your network in the list as well.

Hopefully sorts out #8227

Testing checklist

• Contributor has fully tested the PR manually
• If there are any front-end changes, before/after screenshots are included
• Critical user journeys are covered by Gherkin stories
• Critical and brittle code paths are covered by unit tests

PR process

• PR has the correct target branch and milestone
• PR has 'needs review' or 'work-in-progress' label
• If PR is ready for review, a reviewer has been added. (Don't use 'Assignees')
• If this is an important user-facing change, PR or related issue has a 'changelog' label
• If this includes an internal dependency change, a link to the diff is provided

Reviewer checklist

• Automated test coverage is satisfactory
• PR is fully functional
• PR has been tested for accessibility regressions
• External dependency files were updated if necessary (yarn and pip)
• Documentation is updated
• Contributor is in AUTHORS.md

[pcenov]

Hi @jamalex I tested this build on 2 Windows 10 laptops and now it's not possible to import a facility at all.
Here's a screenshot from the Learner's device showing the disabled radio-button. Notice that there's a /?v=2 parameter added to the the end of the URL which for some reason is added automatically:

Here's a screenshot with the same issue observed when I attempt to import a facility from Device>Facilities:

I'm also observing unusually high CPU usage with this build:

Here are the logs from both laptops:
Admin logs.zip
Learner logs.zip

[jredrejo]

Hi @jamalex I tested this build on 2 Windows 10 laptops and now it's not possible to import a facility at all.
Here's a screenshot from the Learner's device showing the disabled radio-button. Notice that there's a /?v=2 parameter added to the the end of the URL which for some reason is added automatically:

Here's a screenshot with the same issue observed when I attempt to import a facility from Device>Facilities:

I'm also observing unusually high CPU usage with this build:

Here are the logs from both laptops:
Admin logs.zip
Learner logs.zip

Hello @pcenov that's an issue that was solved yesterday. Are you using the build that was created yesterday? https://github.com/learningequality/kolibri/releases/tag/v0.15.0-alpha2 That should have that problem solved.

[pcenov]

Hi @jamalex as suggested by @jredrejo you'll need to rebase develop as he says that the issue I'm facing has been fixed in #8232 and merged yesterday in develop.

[rtibbles]

Hi @pcenov I have rebased this branch now.

[rtibbles]

This feels a little circular, but I approve this PR.

[pcenov]

Hi @rtibbles I did another test run on this using this build and unfortunately the issue described in #8227 is still present.

Here are the logs:
Learner logs.zip
Admin logs.zip

Let me know if you need additional info or anything else.