-
Task
-
Resolution: Fixed
-
Minor
-
None
-
None
-
None
-
-
Sprint 239, Sprint 240, Sprint 241, Sprint 242
The problem is related to proxy sessions (e.g. when mounting a Oneclient). If there is a subscription request and it should be proxied to a different provider, and the target provider is in an incompatible version (or probably offline - check this), sequencer will retry infinitely, without backoffs, and log a lot of error logs. We should:
1. Implement backoffs so that it does not try every 5 seconds. Longest backoff should be e.g. an hour.
2. Limit the logging - if the error is known (e.g. no_connection_to_peer_provider), use a debug log or a throttled warning log.
Below is an excerpt from the logs:
[E 09:43:17.037 <0.22274.235>] Cannot process request
{client_message, {message_id, <<"e42ac1a1e7163b3cc56ab9f9feb33de3ch6cbe">>, <<"573505587337518118">>, <<131,88,100,0,34,111,112,95,119,111,114,107, 101,114,64,102,97,97,115,45,116,101,115,116, 46,100,97,116,97,104,117,98,46,101,103,105, 46,101,117,0,0,86,241,0,0,0,235,100,127,58, 171>>},
undefined,
<<"51994af8ea9515fa1bb64868e9056796ch9515">>,
,
normal,
,
{subscription,<<"-5">>, {file_attr_changed_subscription, <<"Z3VpZCNzcGFjZV9hMmZhYjc2YzYwOGEzNjhkZTU0OGE0YTk5ZDczNzE0NyNhMmZhYjc2YzYwOGEzNjhkZTU0OGE0YTk5ZDczNzE0Nw">>, undefined,500},
undefined}} due to: {badmatch,
{error,
no_connection_to_peer_oneprovider}}. There has been 153 unsuccessful attempts so far. Retrying in 5000 milliseconds...