Adds support for configuration of different types of compression, including none

[chaosbox]

For #6511

This PR adds support for configuration of different types of compression, including none, in the default HTTP filter chain.

• The supported types of compression are those included in go-control-plane/envoy/extensions/compression, namely
  • gzip
  • brotli
  • zstd

Related #310, there had been mentions about disabling compression, the ticket we had raised shows the reason where disabling compression can bring cost benefits.

[github-actions]

Hi @chaosbox! Welcome to our community and thank you for opening your first Pull Request. Someone will review it soon. Thank you for committing to making Contour better. You can also join us on our mailing list and in our channel in the Kubernetes Slack Workspace

[github-actions]

The Contour project currently lacks enough contributors to adequately respond to all PRs.

This bot triages PRs according to the following rules:

• After 14d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, the PR is closed

You can:

• Ensure your PR is passing all CI checks. PRs that are fully green are more likely to be reviewed. If you are having trouble with CI checks, reach out to the #contour channel in the Kubernetes Slack workspace.
• Mark this PR as fresh by commenting or pushing a commit
• Close this PR
• Offer to help out with triage

Please send feedback to the #contour channel in the Kubernetes Slack

[chaosbox]

Ping

[tsaarni]

I've included suggestions inline for some linter nags (link).

Please add the new disableCompression option in "Configuration File" chapter in site/content/docs/main/configuration.md.

Please add also a changelog file.

[geomacy]

Thanks very much for the feedback @tsaarni 🙇 we'll have a look at this and try to get back with updates as soon as possible.

[geomacy]

Many thanks for taking the time to provide the lint fixes. We'll update the docs and add the changelog as soon as possible.

[codecov]

Codecov Report

Attention: Patch coverage is 70.12987% with 23 lines in your changes missing coverage. Please review.

Project coverage is 80.97%. Comparing base (54ceade) to head (45ea946).
Report is 58 commits behind head on main.

Files with missing lines	Patch %	Lines
cmd/contour/servecontext.go	12.50%	13 Missing and 1 partial ⚠️
pkg/config/parameters.go	0.00%	7 Missing and 1 partial ⚠️
cmd/contour/serve.go	50.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #6546      +/-   ##
==========================================
- Coverage   81.76%   80.97%   -0.79%     
==========================================
  Files         133      133              
  Lines       15944    20052    +4108     
==========================================
+ Hits        13037    16238    +3201     
- Misses       2614     3519     +905     
- Partials      293      295       +2     

Files with missing lines	Coverage Δ
internal/envoy/v3/listener.go	98.54% <100.00%> (+0.08%)	⬆️
internal/xdscache/v3/listener.go	92.05% <100.00%> (-0.07%)	⬇️
cmd/contour/serve.go	21.81% <50.00%> (-0.86%)	⬇️
pkg/config/parameters.go	86.00% <0.00%> (-2.03%)	⬇️
cmd/contour/servecontext.go	82.05% <12.50%> (-3.90%)	⬇️

... and 122 files with indirect coverage changes

---- 🚨 Try these New Features:

• Flaky Tests Detection - Detect and resolve failed and flaky tests
• JS Bundle Analysis - Avoid shipping oversized bundles

[geomacy]

Have added one more fix of a test, but I still need to finish testing on-cluster, will update the ticket when I get that completed.

[geomacy]

Hello @davinci26 @tsaarni, I have updated the code per the proposal to support better API extensibility by wrapping the settings in a struct. I believe this PR is ready for review. Some results below from testing on a cluster:

default

out of the box behaviour

curl -v -H "Accept-Encoding: gzip,deflate"  $URL | gzip -d
...
< HTTP/2 200
< content-type: application/json; charset=utf-8
< x-content-type-options: nosniff
< date: Tue, 29 Oct 2024 09:36:04 GMT
< x-envoy-upstream-service-time: 1
< content-encoding: gzip
< vary: Accept-Encoding
< server: envoy
<
{
  "hostname": "podinfo-5bd5b49f6d-n847k",

response was gzip.

parameter disabled

added to container args:

        - --compression=disabled

curl:

$ curl -v -H "Accept-Encoding: gzip,deflate"  $URL
< HTTP/2 200
< content-type: application/json; charset=utf-8
< x-content-type-options: nosniff
< date: Tue, 29 Oct 2024 09:40:55 GMT
< content-length: 353
< x-envoy-upstream-service-time: 1
< server: envoy
<
{
  "hostname": "podinfo-5bd5b49f6d-n847k",

No compression applied, it has been disabled.

parameter gzip

set container args to have

        - --compression=gzip

curl

$ curl -v -H "Accept-Encoding: gzip,deflate"  $URL | gzip -d
...
< HTTP/2 200
< content-type: application/json; charset=utf-8
< x-content-type-options: nosniff
< date: Tue, 29 Oct 2024 09:44:05 GMT
< x-envoy-upstream-service-time: 1
< content-encoding: gzip
< vary: Accept-Encoding
< server: envoy
<
{
  "hostname": "podinfo-5bd5b49f6d-n847k",

parameter brotli

set container args to have

        - --compression=brotli

curl

$ curl -v -H "Accept-Encoding: br,deflate"  $URL | brotli -d
...
< HTTP/2 200
< content-type: application/json; charset=utf-8
< x-content-type-options: nosniff
< date: Tue, 29 Oct 2024 09:46:35 GMT
< x-envoy-upstream-service-time: 2
< content-encoding: br
< vary: Accept-Encoding
< server: envoy
<
{
  "hostname": "podinfo-5bd5b49f6d-n847k",

brotli encoding applied

parameter zstd

set container args with

        - --compression=zstd

curl

$ curl -v -H "Accept-Encoding: zstd,deflate"  $URL | zstd -d
...
< HTTP/2 200
< content-type: application/json; charset=utf-8
< x-content-type-options: nosniff
< date: Tue, 29 Oct 2024 09:49:23 GMT
< x-envoy-upstream-service-time: 1
< content-encoding: zstd
< vary: Accept-Encoding
< server: envoy
<
{
  "hostname": "podinfo-5bd5b49f6d-n847k",

zstd encoding applied

config disabled

Remove compression flag from container args and set config map to have:

apiVersion: v1
data:
  contour.yaml: |-
    compression:
      algorithm: disabled
    ...

Restart contour pods. Do curl requesting gzip

$ curl -v -H "Accept-Encoding: gzip,deflate"  $URL
...
< HTTP/2 200
< content-type: application/json; charset=utf-8
< x-content-type-options: nosniff
< date: Tue, 29 Oct 2024 09:54:23 GMT
< content-length: 353
< x-envoy-upstream-service-time: 1
< server: envoy
<
{
  "hostname": "podinfo-5bd5b49f6d-n847k",
  "version": "6.0.0",

compression is disabled.

config gzip

edit config

apiVersion: v1
data:
  contour.yaml: |-
    compression:
      algorithm: gzip
    ...

Restart contour and curl:

$ curl -v -H "Accept-Encoding: gzip,deflate"  $URL | gzip -d
...
< HTTP/2 200
< content-type: application/json; charset=utf-8
< x-content-type-options: nosniff
< date: Tue, 29 Oct 2024 09:56:37 GMT
< x-envoy-upstream-service-time: 1
< content-encoding: gzip
< vary: Accept-Encoding
< server: envoy
<
{
  "hostname": "podinfo-5bd5b49f6d-n847k",

response encoded with gzip

config brotli

edit config to

apiVersion: v1
data:
  contour.yaml: |-
    compression:
      algorithm: brotli

Restart contour and curl:

$ curl -v -H "Accept-Encoding: br,deflate"  $URL | brotli -d
...
< HTTP/2 200
< content-type: application/json; charset=utf-8
< x-content-type-options: nosniff
< date: Tue, 29 Oct 2024 09:59:48 GMT
< x-envoy-upstream-service-time: 1
< content-encoding: br
< vary: Accept-Encoding
< server: envoy
{
  "hostname": "podinfo-5bd5b49f6d-n847k",

config zstd

edit config to

apiVersion: v1
data:
  contour.yaml: |-
    compression:
      algorithm: zstd

curl:

$ curl -v -H "Accept-Encoding: zstd,deflate"  $URL | zstd -d
...
< HTTP/2 200
< content-type: application/json; charset=utf-8
< x-content-type-options: nosniff
< date: Tue, 29 Oct 2024 10:29:20 GMT
< x-envoy-upstream-service-time: 1
< content-encoding: zstd
< vary: Accept-Encoding
< server: envoy
<
{
  "hostname": "podinfo-5bd5b49f6d-n847k",

zstd compression applied.

[geomacy]

Hello @tsaarni, @davinci26, would you be able to review this? Thanks.

[davinci26]

@geomacy will take a look later today.

[geomacy]

hi @davinci26 @tsaarni did you get a chance to look at this?

[tsaarni]

Apologies for the long delay! I've finally had a chance to review and added comments inline.

The PR title and description are slightly outdated after the latest changes, and could use an update.

Thank you for your efforts and patience!

[tsaarni]

Let's remove the command line option and make configuration available only through config file and ContourConfiguration CRD.

The command line options currently are a bit of a mess, so we previously agreed to avoid introducing new options there unless absolutely necessary.

[geomacy]

See 95d38af

[tsaarni]

Would it be cleaner to treat an empty string as a valid value here, rather than completely bypassing validation in Parameters.Validate()? While algorithm is the only field currently, that might change.

Because +optionalstring in Go becomes an empty string when user did not set the field, it might be best to consider it as valid algorithm name in the Go side (CRD validation for the field can still reject it). It is already taken into account elsewhere in this change: inhttpConnectionManagerBuilder.DefaultFilters() you've added switch-default branch which will use gzip for empty algorithm string.

[geomacy]

See 366f48d

[tsaarni]

It seems that DefaultFilters() function used to set up a set of fixed filters - I don't see any of the other builder method impacting the default set except the new Compression() method, which now also makes it important in which order the builder methods are called - Now the configuration would be different depending if Compression() is called before or after DefaultFilters().

Maybe it would be better to (1) remove compressor from the default set completely and only add it when the new Compression() builder method is called? Or maybe some other alternative: (2) instantiate gzip by default and modify it later according to chosen algorithm when Get() is called or (3) delay adding compressor until Get() but add it by default at that point.

[geomacy]

It's true that the way the code is at present means the order does matter. At the same time the intent of the PR is to allow a way of changing the compression algorithm in the default filter chain (as opposed to something more generic), so it kind of makes sense that DefaultFilters() no longer sets up a fixed set. Maybe Compression() is a bad name for that method, perhaps it should be something like OverrideDefaultCompression(), and documented more clearly to explain the behaviour including how it needs to be called before DefaultFilters() if it is to do anything.

I've been trying to avoid changing the code in any way that would risk changing behaviour for users who don't opt in to the new method. At a basic level this certainly means we can't change the signatures say of HTTPConnectionManagerBuilder() or DefaultFilters() itself, but I would also want to avoid changing the behaviour of DefaultFilters() in the absence of using the new compression method. So in thinking about these options:

About 1. Remove compressor from the default set entirely.
I wouldn't be keen to do this as the behaviour of adding compression is effectively already part of the contract of what DefaultFilters() does. Changing this could change the behaviour of downstream users' code.

Option 3 is similar, in that you're changing the order of filters (currently default filters put compression filter first). Less dramatic than option 1 but still a change in behaviour.

Option 2 might be possible - I guess Get() could scan the filter chain looking for one named "envoy.filters.http.compressor" (value of CompressorFilterName as set in DefaultFilters() at present) to update. Would that be good enough? Can it update the first such filter it finds, on the assumption that there will be only one?

I'd like to be sure about any changes before I make them here, but my feeling is that perhaps no change is needed - other than renaming the Compression() func and providing better documentation. What do you think?

[tsaarni]

Compression here is still string while it should be CompressionAlgorithm. The new type needs a new table with algorithm field with the content described here.

[geomacy]

see 1a6465f

[geomacy]

Many thanks for the comments @tsaarni will have a look through and apply changes as soon as possible.

[geomacy]

hi @tsaarni, thanks for the comments and suggestions. I have a branch in preparation to do these and will merge it here when I see tests pass. [Update: changes added now].

What do you think about the notes here on the question about the builder methods?

[geomacy]

hi @tsaarni what do you think about the above changes and question?

[geomacy]

hi @tsaarni I have thought about the above further and I do feel that the clarification of the method's usage is the best approach - I have added this in 2f5e157.

Please have another look at the latest batch of commits, thanks.

[geomacy]

@tsaarni any thoughts please?

[geomacy]

hi @tsaarni, @davinci26, @sunjayBhatia we would really appreciate it if you could get back to us on this pull request, which has been sitting for some weeks. I believe from the above conversations that this is either ready for final approval and merge, or very nearly so. The changes on this PR have been going on for a long time and it would be really good to get it finished off at last. Thanks!