The logs are currently inconsistent, the messages use different casing and tenses, and some include valuable fields and others do not. There is some information we should always be able to figure out looking at the debug logs...

• image name for a snapshot after a pull happens
• snapshot operation failures
• metadata operation failures

Likewise there is probably some logs being logged at the error level that we only want for debugging. We should also be careful to avoid using the phrase failure when an operation has not failed (such as not exists).