Arvados

11309-crunchstat-summary-containers @ dc140f2c277a7571669f5bdf492bac6448a0fba0

I added "--container UUID" and "--container-request UUID" options to make it more obvious, but under the hood they're the same as "--job": they figure out the object type from the UUID.

I also dealt with #10359 because it was easy: added a --threads N option, default 8.

The report is unwieldy for a container with lots of "child" containers. Might be better to sort them by name (or creation time of container?) instead of UUID.

11309-crunchstat-summary-containers @ f04ff34e47cc5ff1e625e9a5c1b1b0cbeb810d8c

Child containers are now sorted by creation time of container request.

A couple comments:

• Failing test: test_container (tests.test_examples.SummarizeContainer)
• File tools/crunchstat-summary/crunchstat_summary/summarizer.py
  • Line 437: Is there a case where process could be a dict? If so, what happens with the uuid variable?

Lucas Di Pentima wrote:

• Failing test: test_container (tests.test_examples.SummarizeContainer)

Oops, updated fake API record to include created_at.

• File tools/crunchstat-summary/crunchstat_summary/summarizer.py
  • Line 437: Is there a case where process could be a dict? If so, what happens with the uuid variable?

This "pass already-retrieved object instead of just uuid" didn't end up getting used here but I was thinking it might end up being useful in #10472. But yes, it was broken; I've fixed it, and changed the arg to "process_or_uuid" to make it readable.

11309-crunchstat-summary-containers @ c77cb0091a8488eae52341920e8e32beb8fc47dc

This LGTM, thanks.