and_recategorization.csv and and_scope.py don't appear to be reflected clearly on http://phlogiston.wmflabs.org/and_report.html
Some images are failing to load, and "These charts include data from the following Phabricator projects, in this order" is not reflected from and_scope.py
Forecasts images are show generic image placeholders, and Project Burnup by count is blank/white on the page.
Images are failing to load due to this R error:
Error: Aesthetics must be either length 1 or the same as the data (1): x, y, label
I think this tends to reflect having empty data. Checking the database:
phab=# select category, count(*) from task_history_recat where scope = 'and' group by category; category | count ----------+-------- Other | 310535 (1 row)
So the charts are empty because there is only one category of data and it's a hidden category.
There's only one category of data (Other) because the other category matches on `%Android-app-feature-Feeds%`, which isn't present in the raw data.
Not sure why the unzoomed charts are all broken as well, though, since "other" should still be present in unzoomed charts. Will investigate further.
The missing unzoomed burnup and tranche charts are a problem for all projects, not just Col. But having the non-zoomed charts be missing for And is due to the configuration as described above.
T138963: Tranche and unzoomed charts are broken for all projects.
@JAufrecht I get a 502 Bad Gateway error when trying to access http://phlogiston.wmflabs.org/
phlogiston-dev.wmflabs.org/and_category_possibilities.txt also times out.
Where can I edit the possibilities, assuming everything is working?
Phlogiston-1 is currently up and running, and in the middle of a fresh Phlogiston reporting run, which should take a few hours.
Phlogiston-2 is currently up and running, and also in the middle of a fresh run. You should be able to view category_possibilities on either server. I'll keep an eye on it (you can also watch the timestamps change sequentially on the front page) and let you know then you can run. Also, the batch is running in the mission_control tmux instance, so that's a clear mechanism to avoid conflicting commands.
See T137736: Phlogiston-1 unresponsive and T129891: phlogiston-2 hangs every week for more background on outages.