department-of-veterans-affairs / va.gov-cms

Editor-centered management for Veteran-centered content.
https://prod.cms.va.gov
GNU General Public License v2.0
98 stars 68 forks source link

Events v2 optimization - content build slowdown due to code changes #9440

Closed brianalloyd closed 2 years ago

brianalloyd commented 2 years ago

Description

Around Jan 3 2022, the average content build time became approximately 5 minutes slower: https://app.datadoghq.com/dashboard/eb9-ece-f9z/platform-cms-prod-cms?fullscreen_end_ts=1642813745148&fullscreen_paused=false&fullscreen_section=overview&fullscreen_start_ts=1640135345148&fullscreen_widget=7727346438800838&from_ts=1640369100921&to_ts=1643047500921&live=true

There does appear to be some change in GraphQL time at this point, but it does not explain the shift. Taking GraphQL time out of total build time shows that the non-GraphQL time has also jumped: https://app.datadoghq.com/dashboard/gm3-xnt-et5?fullscreen_end_ts=1643047588779&fullscreen_paused=false&fullscreen_section=overview&fullscreen_start_ts=1640369188779&fullscreen_widget=8055303753625958&from_ts=1640369184121&to_ts=1643047584121&live=true

The jump almost certainly happened as we came out of code freeze. Therefore, there is a range of work that could be responsible for the change. This is the full range of commits from 12/23 - 1/5, beginning of code freeze until just beyond: https://github.com/department-of-veterans-affairs/content-build/compare/22763456b64f001903126c2cebe2a38b0fe8c561...4b29aa4bb7405a01fb25255811d545da9fe2d934

We should do timeboxed investigation into this to identify the changes that have caused the slowdown. If fixing the problem is more involved than an hour's work, a follow up ticket should be created with the team which owns the code in question to rectify the problem.

Considerations

Frontend Tasks

Acceptance Criteria

jilladams commented 2 years ago

@ethanteague and @maxx1128 this is an old ticket but I'm curious if this came up/ was addressed during your recent work. (cc @timcosgrove in case you have any recent notes since #8801 )

ethanteague commented 2 years ago

@ethanteague and @maxx1128 this is an old ticket but I'm curious if this came up/ was addressed during your recent work. (cc @timcosgrove in case you have any recent notes since #8801 )

No, it sure didn't / wasn't on our radar, @jilladams. That said, we now have a pattern we can follow = KISS architecture based on Covid status endpoint file that is accessible without SOCKS.

jilladams commented 2 years ago

Closing: https://dsva.slack.com/archives/C52CL1PKQ/p1659566857747289?thread_ts=1659545826.901579&cid=C52CL1PKQ