tools for digging into CQ cycle times?

[Ojan Vafai]

https://trooper-o-matic.appspot.com/cq/chromium is awesome data, but I'm not sure what to do next in understanding where the time is going. Do we have a reasonable way of digging in and seeing what caused 99th percentile to be 436 minutes at one point?

Something like https://trooper-o-matic.appspot.com/tree/chromium would be awesome, but it's only for the main waterfall at the moment, right?

[Eric Seidel]

chromium-cq-status.appspot.com likely has all the data you need. Just
needs to be reduced into the number you want.

> --
> You received this message because you are subscribed to the Google Groups
> "Chromium Hackability Code Yellow" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to hackability-c...@chromium.org.
> To post to this group, send email to hackabi...@chromium.org.
> To view this discussion on the web visit
> https://groups.google.com/a/chromium.org/d/msgid/hackability-cy/CANMdWTvztc4CLLrOU6_cYD646Mc%2BsGyxJk3bYFz%2BARsY7_ZygQ%40mail.gmail.com.

[Eric Seidel]

Alan or Sergey are the people to talk to about chromium-cq-status.

[John Abd-El-Malek]

btw the tracking bug is https://code.google.com/p/chromium/issues/detail?id=397168

[Ojan Vafai]

Yup. That seems promising. What I really want is to be able to click on a column of https://trooper-o-matic.appspot.com/cq/chromium and get to the list of try runs in that column sorted by time (i.e. the slowest runs at the top). I expect this is just forming the right query to http://chromium-cq-status.appspot.com/, but it's hard for me to tell.

[Sergey Berezin]

The data you likely want here is how much time each CL took in each of the CQ stages (waiting for CQ to start working on it, waiting for try jobs to run, waiting for tree to open, waiting for throttling in CQ, waiting for the actual commit, etc.). The raw data is available in chromium-cq-status, but its aggregation still needs to happen outside. I'll update http://crbug.com/397168 with more detail.

There is no magic query to chromium-cq-status to get a list of the longest CLs, but again, the length of each CL can be derived from the logged data and presented by trooper-o-matic or a stand-alone python script. I think we can start with a stand-alone script and port it to trooper-o-matic. Details will be in the bug.

Sergey.

[Julie Parent]

+eae (as an FYI of explicit data that needs better collection, storage, and analysis tooling)

Emil is working up a proposal for a centralized robust data store and query api for assorted infra team data.  This is clearly data that should be going into that system.