Open hickeng opened 7 years ago
As a first iteration, option (2.) is the easiest and will satisfy customer requirements for now. Basically, the syslog endpoint should be routable either through the public network via the VCH or by another network the container is connected to. This also means that containers using --net=none
will not have their logs redirected. The same endpoint specified in #4861 can be used for container logs as well.
Story As a user I want to have my containerized application logs processed and aggregated at a remote collector
Detail Broken out from #4613 (this comment and previous)
Using loginsight, splunk or similar to process logs can add significant value. docker-engine provides the logging driver mechanism which can be used to accomplish this style of redirection. With the VCH/container VM its logs are written directly to the datastore to avoid loss in the case of the endpointVM being down, intentionally or otherwise.
This requires some other means of redirecting logs to a logging service. Some options: