We've received reports that users can't connect to machines which are
experiencing a high number of network change events. It looks like
this is because the FtlSignalStrategy class is signing the user out
and back in for every network error.
The FtlRegistrationManager has its own logic for updating the FTL
auth token when it is about to expire and handles network errors
independently so we don't need to explicitly sign the user out for
retriable network errors.
One concern would be whether fixing this will result in a different
service API returning RESOURCE_EXHAUSTED since the host will
need to re-establish a signaling channel and report both offline
and online status to our backend. I tested this change using a
timer which disconnected the signaling channel every 30 seconds for
several hours and did not encounter any issues with signaling.
Bug: b:329298701
Change-Id: Ifd92cf732b50cf6ee264c5b87cef13ddb506233a
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5393341
Reviewed-by: Jamie Walch <jamiewalch@chromium.org>
Commit-Queue: Joe Downing <joedow@chromium.org>
Cr-Commit-Position: refs/heads/main@{#1277798}
47db858b4a