Jamil
Jamil
> > > @AndrewDryga I just realized that now we can get some filters with messages "all", does it still mean that empty filters means allow all or allow none?...
This won't work because we have only Layer 3 information to identify the source traffic for reversing the external IP back into the proxied IP to send data back to...
I think this also happens when the portal and/or relay is restarted while the mapping is in place? Is that possible?
I think this is what's causing #4521 -- if DNS mappings fail to be refreshed before TURN credentials expire that would explain this behavior
100.64.0.1 -> 10.10.10.10 -> gitlab.company.com 100.64.0.1 -> 10.10.10.10 -> jira.company.com 100.64.0.2 -> 10.10.10.10 foo.company.com 100.64.0.1 -> -> new gateway: resolve (gitlab.company.com, jira.company.com): new_mappings (10.10.10.10) First priority: - Don't drop connections...
@conectado Yeah I think we can maybe save this issue for the explicit case for handling Gateway restarts, which will kill the application-layer connection anyway, so DNS will be re-queried....
> Not neccesarily, if an application such as a browser caches the DNS queries then it wouldn't re-query the DNS Yeah, we may just need to communicate this to admins...
This should be considered when implementing #3343
There won't be new `init` messages, only DNS config and other updates.
@roop I suspect fixing this for your iOS 15 will also fix it for macOS 12. It seems like the container store is not getting set up properly for writes...