Check probe internet connection on "Connection to API failed"

[alexey-yarmosh]

Inside the connectError handler we can run the status-managers's pingTest, to confirm or deny whether the problem is on the probe network side.

Benefits:

• If a failure of a probe network is confirmed we can write a human-readable log for users, informing them that there is a problem with their network.
• It will give extra metadata during debug of unusual connectivity problems.

[MartinKolarik]

Note that the current QA is going to be reworked to test just our API (#323). We can consider this after, but I think it's relatively low-value. For us, a good indicator of probe vs API issues is the stats of connected probes over time, which we already have, and if the connection fails randomly, or just for some sites, it is still an issue with the network.

[alexey-yarmosh]

Note that the current QA is going to be reworked to test just our API (#323).

Ok, this can be a separate check, not from status-manager.

For us, a good indicator of probe vs API issues is the stats of connected probes over time, which we already have, and if the connection fails randomly, or just for some sites, it is still an issue with the network.

N of connected probes over time shows it in general, but I guess it is possible to have a specific problem, e.g. GP IP is blocked in some country.

[jimaek]

yeah it doesnt have to be part of the QA check, just something to prove to users and allow us to verify, if the API is really inaccessible or the probe's internet is the issue. Currently when checking some logs you might think our API is constantly going down and users dont find that reinsiruing.