PR#162: get_bz_query: use last_update_string for bug creation too - fedora-qa/blockerbugs

fedora-qa / blockerbugs

#162 get_bz_query: use last_update_string for bug creation too

Merged 4 years ago by kparal. Opened 4 years ago by adamwill.

fedora-qa/ adamwill/blockerbugs bug-creation-ts into develop

get_bz_query: use last_update_string for bug creation too

Adam Williamson • 4 years ago

84f25e4

blockerbugs/util/bz_interface.py

file modified

+1 -7

		`@@ -76,17 +76,11 @@`
		`query['keywords'] = 'Triaged'`
		`if last_update:`
		`last_update_string = last_update.strftime("%Y-%m-%d %H:%M GMT")`
		`- yesterday_timestring = (last_update - datetime.timedelta(days=1)).strftime("%Y-%m-%d")`

		`# Since this is kind of confusing, the idea here is to only grab only`
		`# the bugs whose whiteboard, blocks or dependson has changed since`
		`# the last sync or any bugs that have been created in the last day.`

		`- # Unfortunately, there doesn't seem to be a way to filter the created`
		`- # bugs in a way that's more granular than a day, so we'll pull in`
		`- # extra bugs and slow down the sync process but the alternative is`
		`- # missing bugs from the list`
		`-`
		`# keep in mind that OP -> open parenthesis, CP -> closed parenthesis`
		`# and the OR is to make sure that a bug is returned if ANY of the`
		`# changed after conditions are met`
		`@@ -107,7 +101,7 @@`
		`'v6': last_update_string,`
		`'f7': 'creation_ts',`
		`'o7': 'greaterthaneq',`
		`- 'v7': yesterday_timestring,`
		`+ 'v7': last_update_string,`
		`'f8': 'short_desc',`
		`'o8': 'changedafter',`
		`'v8': last_update_string,`

adamwill commented 4 years ago

The comment from @tflink saying this doesn't work, which this
commit removes, seems to be no longer valid. I tested using a
query for creation_ts greaterthaneq "2020-11-27" and then making
it "2020-11-27 16:00 GMT"; the former gave 202 results, the
latter gave 101. That looks a lot like it works.

Signed-off-by: Adam Williamson awilliam@redhat.com

adamwill commented 4 years ago

Review from @tflink would be appreciated if he remembers this at all :) This is the test I did:

>>> import bugzilla
>>> bz = bugzilla.RHBugzilla4("https://bugzilla.redhat.com/xmlrpc.cgi")
>>> query = {'f1': 'OP', 'f2': 'creation_ts', 'o2': 'greaterthaneq', 'v2': "2020-11-27", 'f3': 'CP'}
>>> len(bz.query(query))
202
>>> query = {'f1': 'OP', 'f2': 'creation_ts', 'o2': 'greaterthaneq', 'v2': "2020-11-27 16:00 GMT", 'f3': 'CP'}
>>> len(bz.query(query))
101

I wrote this after looking at the timeouts we're getting on sync (see emails to sysadmin-qa), but it will only help if we're not doing a full sync. I'm not sure off the top of my head whether we are.

zuul commented 4 years ago

Build succeeded.

blockerbugs-pytest : SUCCESS in 1m 28s

kparal commented 4 years ago

The diff looks good on paper.

@tflink Any thoughts about this?

tflink commented 4 years ago

It's been 8-ish years since that code was written, glad to see that bugzilla has improved in this way.

If the code works, I have no objections. I think that it did something obviously wrong when I used to use the whole timestamp, the test that @adamwill did shows that's no longer the case.

We should only do full syncs when forced or when initializing the database. I don't think that has changed, anyways.

Either way, +1 from me

kparal commented 4 years ago

Commit 924266d fixes this pull-request