Experiment with not caching the ~/.npm directory during Filesize/Tests CircleCI jobs
#10209
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Looking at a recent job, restoring and later saving the 400MB+ npm cache takes 13 and 14 seconds, respectively: https://app.circleci.com/pipelines/github/apollographql/apollo-client/17326/workflows/4db5bc1c-718e-4f0b-b6ec-7089b9aa8b55/jobs/94670 This job took a total of 1 minute 11 seconds, so restoring/saving the cache accounts for ~38% of the total time. If removing the caching altogether does not slow anything else down by that much (27 total seconds), then the caching was not really delivering on its intended goals, and should be removed. Update after running CI on this commit: https://app.circleci.com/pipelines/github/apollographql/apollo-client/17327/workflows/0c21e96a-736e-48f2-b2a9-55dea5405342/jobs/94672 Now only 1 minute 1 second (noticeably less time without any caching)! The "Spin up environment" step went from taking 2s to 19s, because the circleci/node@16.13.1 Docker image needed to be re-pulled. This seems like something that could be addressed separately, if we care. The `npm ci` step takes a little longer, 9s instead of 4s, but that's presumably because everything has to be installed without a cache. Still, this is much faster than restoring `node_modules` from cache.
|
Good news: the pulling of the |
benjamn
changed the title
~/.npm directory during Filesize/Tests CircleCI jobs
benjamn
commented
Oct 19, 2022
.circleci/config.yml
Outdated
Comment on lines
10
to
16
| # - restore_cache: | ||
| # keys: | ||
| # # When lock file changes, use increasingly general patterns to | ||
| # # restore cache | ||
| # - npm-v3-{{ .Branch }}-{{ checksum "package-lock.json" }} | ||
| # - npm-v3-{{ .Branch }}- | ||
| # - npm-v3- |
There was a problem hiding this comment.
Do we want to leave these sections commented out for future reference, or just remove the lines and let Git history be our memory?
There was a problem hiding this comment.
I generally prefer git history, but not a strong opinion.
alessbell
approved these changes
Oct 19, 2022
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Looking at a recent job, restoring and later saving the cached
~/.npmdirectory takes 13 and 14 seconds, respectively. This job took a total of 1 minute 11 seconds, so restoring/saving the (large)~/.npmdirectory accounts for ~38% of the total time.If removing the caching altogether does not slow anything else down by that much (~27 total seconds), then the caching was not really delivering on its intended goals, and should be removed.
New results after running CI on this commit/PR. Now only 1 minute 1 second (noticeably less time without any caching)!
Notes:
The "Spin up environment" step went from taking 2s to 19s, because the circleci/node@16.13.1 Docker image needed to be re-pulled. This seems like something that could be addressed separately, if we care.
The
npm cistep takes a little longer, 9s instead of 4s, but that's presumably because everything has to be installed without a cache. Still, this is much faster than restoring~/.npmfrom cache.By itself, this PR does not (yet) speed up the finishing time of CI checks, since running the tests (the other CI job) always takes longer than running the Filesize job does now (or did before).