[Performance] MiqEventDefinition.seed speedup

[kbrock]

MiqEventDefinition is the slowest of the primordial seeds

The main culprit is the fact that miq_sets uses relationships.

ms	bytes	objects	query	query ms	rows	comments
2,595.0	51,390,823*	645,934	719	1,653.5	1,036	nop before
2,121.1	45,063,997*	577,455	542	1,297.0	1,036	nop after
18%	12%	11%	25%	22%	0	diff

details

This removes a number of N+1 queries but does not remove the main culprit. That would probably require reimplementing miq_sets.

If you note the SELECT miq_event_definitions are merged together. 338ms+36ms => 51ms

ms	bytes	objects	query	query ms	rows	comments
2,595.0	51,390,823*	645,934	719	1,653.5	1,036	MiqEventDefinition-1
338.8			178	305.9	178	.SELECT "miq_event_definitions".*
36.0			14	29.1	166	.SELECT "miq_event_definitions".*
628.8			167	579.5	180	.SELECT "miq_sets".*
190.2			166	162.7	166	.SELECT "relationships"."ancestry"
459.7			194	434.6	346	.SELECT "relationships".*

ms	bytes	objects	query	query ms	rows	comments
2,121.1	45,063,997*	577,455	542	1,297.0	1,036	MiqEventDefinition-1
51.5			15	31.1	344	.SELECT "miq_event_definitions".*
569.9			167	523.2	180	.SELECT "miq_sets".*
196.6			166	167.8	166	.SELECT "relationships"."ancestry"
479.0			194	455.8	346	.SELECT "relationships".*

related to:

• https://bugzilla.redhat.com/show_bug.cgi?id=1422671
• [Performance] MiqGroup.seed #15586

[carbonin]

Looks good to me, just fix up the variable shadowing.

The two e's are a bit hard to understand.

[miq-bot]

Checked commit kbrock@8013df3 with ruby 2.3.3, rubocop 0.47.1, and haml-lint 0.20.0
3 files checked, 0 offenses detected
Everything looks fine. ⭐

[kbrock]

Thanks @carbonin

Changes:

• removed the e shadow.
• The events created are now added to the local cache of events. Duplicates in the file (which shouldn't happen) will behave as before.