GitLab

Provide several fixes and improvements for eNB log forwarding, targeting in particular the default forwarding to Wendelin Telecom:

• Fix the generation of the fluentbit tag, where the default value for enb_id and gnb_id is now "" instead of None, which produced malformed tags like ors123_COMP-1234_eNone instead of ors123_COMP-1234.
• Add a boolean instance parameter xlog_forwarding_enabled (True by default) that activates or deactivates log forwarding through fluentbit entirely. Specifically, when set to False, the sections pertaining to fluentbit are removed, and the Wendelin Telecom shared instance's destruction is requested.
• Remove the default value for xlog_fluentbit_forward_host, and only forward logs (if xlog_forwarding_enabled == True) if the parameter is explicitly set OR the Wendelin Telecom shared instance returns the forward host address once it has been correctly requested. Similarly to above, the Wendelin Telecom shared instance is destroyed if the parameter is left empty (i.e. the default behavior).

Currently, when xlog_fluentbit_forward_host is unset, it produces a faulty line in the fluentbit config:

    Host

which crashes the fluentbit service until the parameter is provided. This only happens in the case where we want to forward the logs to Wendelin Telecom (xlog_forwarding_enabled == True and xlog_fluentbit_forward_host == ""): I consider it reasonable to flag the problem in this explicit way until the desired parameter is received from the shared instance.

• Add two simple promises that monitor the fluentbit instance, and to make sure the partition is reprocessed until it runs correctly:

1. check-xlog-fluentbit-forward-host checks that the forward host and port combination is valid by attempting a connection through a socket.
2. check-xlog-fluentbit-health uses fluentbit's Monitoring features by querying the /api/v1/health of a dedicated monitoring HTTP server enabled in the fluentbit configuration. If the service is healthy, the server returns a 200 status code; it returns a 500 status code when fluentbit starts losing data chunks due to errors or not being able to retry forwarding them.