There are a number of silent failures in various services (sync, indexing, etc) that are logged but otherwise tend to go unnoticed until a later date. While we continue working to make services more robust and add instrumentation where needed, it seems to me we can set up some form of proactive monitoring and alerting system on our logs to catch known issues sooner.
We agreed to start by keeping a running list of log output lines for known silent failures. Eventually, the list could be used to set up said monitoring and alerting system.
There are a number of silent failures in various services (sync, indexing, etc) that are logged but otherwise tend to go unnoticed until a later date. While we continue working to make services more robust and add instrumentation where needed, it seems to me we can set up some form of proactive monitoring and alerting system on our logs to catch known issues sooner.
We agreed to start by keeping a running list of log output lines for known silent failures. Eventually, the list could be used to set up said monitoring and alerting system.