Proc stats

[tj]

Potentially collect and expose metrics from the server process and report to cloud watch, data dog, etc much like the logging integrations. These aren't nearly as relevant in the FaaS land, as outliers causing memory issues are isolated and so on, but you may still have leaks which cause the function to exceed the limits.

• provide "light" and "extended" variants (ex general CPU usage vs user/system/iowait/etc)
• CPU stats
• Mem stats
• Disk stats
• Load avg
• Net stats

[stephenmathieson]

I may be off, but I think all of this is already available (and pretty easy to find) in the AWS console. Is the idea just to put some polish on top of what already exists, or to collect separate (but duplicate?) metrics?

[tj]

Lambda only provides invocation count / duration / errors / throttles at the moment, not even memory which is odd considering it's one of the limits haha.. could be real close to hitting the function limit and not know

[stephenmathieson]

You're totally right, my bad. +1 here then ;)