tracking vs logging

[arielleleon]

There was an open PR where the term "Logging" was used to indicate the documenting of all things that happen in the lab management system. It would be good for us to all define how we use the logging term so we use a shared vocabulary

[patricklatimer]

In SIPE's effort to get good at observability, we've learned a lot about good observability practices and standards, and there tends to be an industry standard agreement that an "Observability" or "Telemetry" system emcompasses the following three things:

• Logs: messages from applications
• Metrics: system monitoring (e.g. CPU)
• Traces:

(@jessy-liao feel free to chime in or correct me)

A principle that we should really strive to uphold is that observability tools should be a one-way street, and should not be part of a closed-loop control system. That is, we shouldn't be using application logs to tell us when an experiment or a process is finished. There should be a different system for that.

Note

It's fine (and probably smart) to collect the logs/metrics/traces from an experimental session and deliver it with the data asset, that way if you're analyzing the data and something is fishy, you can easily see if, for example, the CPU was maxing out on the experimental computer. But at this point it's kind of just a different kind of metadata or data

So it's good we're building that system into the LIMS! I just want to make sure we think about a good term we can use for that functionality that differentiates it from the observability platform. Some options off the top of my head:

• Tracking
• History
• Records
• Execution Archive

[dougollerenshaw]

@arielleleon who should this be assigned to?