Get deployment back on track w/ npm in search

[chadwhitacre]

When we deployed #4458 yesterday, the site slowed down significantly, to the point where it was unusable and we rolled back in Heroku (slack). We need to get back on track.

[rohitpaulk]

Health dashboard for the last 3 days:

[rohitpaulk]

The response time spike started at 9 May, 18:30 (UTC) (around midnight indian time).

I see 2 deploys around that time:

[rohitpaulk]

The first one in that screenshot (from the bottom) is: 64ff91e

[chadwhitacre]

I rolled us forward again a few minutes ago and the site is still exhibiting normal response times relative to the (higher perc99) baseline since #4305.

[chadwhitacre]

@rohitpaulk I think we should close this out and open a new ticket about bringing our overall response times down. Eh?