compressor: add weaken_etag_on_compress to weaken strong ETags when compressing by rafaelgaspar · Pull Request #43755 · envoyproxy/envoy

rafaelgaspar · 2026-03-04T11:35:51Z

Generative AI Disclaimer: This PR was prepared with assistance from generative AI (Cursor). The implementation, tests, and documentation have been fully reviewed by me as the PR author, and I take responsibility for the changes.

Commit Message:
compressor: add weaken_etag_on_compress to weaken strong ETags when compressing

Additional Description:
Adds a new option weaken_etag_on_compress to the compressor filter's
response_direction_config. When enabled, strong ETag response headers
are weakened by prepending W/ to the value (e.g. "abc123" becomes
W/"abc123") instead of being removed when compression is applied. Weak
ETags (already starting with W/) are left unchanged.

This allows caches and conditional requests to keep working while indicating
that the representation was modified by compression, matching behavior used in
other proxies (e.g. Varnish). The default remains false, preserving
existing behavior (strong ETags are removed when compressing).

Fixes #35152

Risk Level: Low

Testing:
New unit tests in compressor_filter_test.cc:

WeakenEtagOnCompressStrongEtag – strong ETag is weakened when compressing
WeakenEtagOnCompressStrongEtagWithQuotes – strong ETag with quotes is weakened
WeakenEtagOnCompressWeakEtagUnchanged – weak ETag is preserved
WeakenEtagOnCompressWithStatusHeaderEnabled – weaken behavior with status header enabled

Docs Changes: Yes

Updated docs/root/configuration/http/http_filters/compressor_filter.rst with an "ETag handling" section describing disable_on_etag_header and weaken_etag_on_compress, and added a YAML example for weaken_etag_on_compress.
Updated "When response compression is applied" and compression-skip conditions to reference ETag handling.
Clarified EtagNotAllowed status and not_compressed_etag stat descriptions.

Release Notes: Yes

Added entry under new_features in changelogs/current.yaml for the compressor filter's weaken_etag_on_compress option.

Platform Specific Features: N/A

repokitteh-read-only · 2026-03-04T11:35:59Z

Hi @rafaelgaspar, welcome and thank you for your contribution.

We will try to review your Pull Request as quickly as possible.

In the meantime, please take a look at the contribution guidelines if you have not done so already.

🐱

Caused by: #43755 was opened by rafaelgaspar.

see: more, trace.

repokitteh-read-only · 2026-03-04T11:36:05Z

CC @envoyproxy/api-shepherds: Your approval is needed for changes made to (api/envoy/|docs/root/api-docs/).
envoyproxy/api-shepherds assignee is @adisuissa
CC @envoyproxy/api-watchers: FYI only for changes made to (api/envoy/|docs/root/api-docs/).

🐱

Caused by: #43755 was opened by rafaelgaspar.

see: more, trace.

agrawroh · 2026-03-10T18:32:26Z

cc @adisuissa Gentle ping for the review.

agrawroh · 2026-03-10T18:32:36Z

@rafaelgaspar It also needs a main merge.

/wait

adisuissa

Thanks! left a couple of high-level API comments.

rafaelgaspar · 2026-03-10T22:36:08Z

@adisuissa both suggestions have been incorporated into the code, let me know if there is anything else.

paul-r-gall · 2026-03-16T13:53:04Z

@adisuissa re-ping for "stalled" bit.

adisuissa

/lgtm api

adisuissa · 2026-03-17T20:20:41Z

This requires a compressor filter owner review.
/assign @KBaichoo

KBaichoo

looks good otherwise, thanks

KBaichoo · 2026-03-20T16:40:28Z

/wait

Introduces a new configuration option, `weaken_etag_on_compress`, to the compressor filter's response direction configuration. When enabled, strong ETag headers are modified to include a "W/" prefix instead of being removed during compression, allowing for better cache and conditional request handling. Updates to documentation and tests have been made to reflect this new behavior. Signed-off-by: Rafael Antunes <rafael.gaspar@me.com>

Updated the compressor filter to clarify the precedence of the `weaken_etag_on_compress` option when both it and `disable_on_etag_header` are enabled. Compression will now be applied, and the ETag will be weakened, allowing for better cache management. Documentation and tests have been updated to reflect this behavior. Signed-off-by: Rafael Antunes <rafael.gaspar@me.com>

Signed-off-by: Rafael Antunes <rafael.gaspar@me.com>

Introduced a helper function `isWeakEtag` to streamline the logic for identifying weak ETags in the `sanitizeEtagHeader` and `weakenEtagHeader` methods. Signed-off-by: Rafael Antunes <rafael.gaspar@me.com>

Updated the initialization of the `weaken_etag_on_compress` member variable to directly use the value from the response direction configuration, simplifying the logic and ensuring consistent behavior. Signed-off-by: Rafael Antunes <rafael.gaspar@me.com>

rafaelgaspar · 2026-03-23T10:10:21Z

@KBaichoo addressed the comments, and did a rebase. Let me know if there is anything else.

KBaichoo

/wait

Updated the CompressorFilter to remove strong ETags of length two during compression, aligning their treatment with longer strong ETags. Documentation has been updated to reflect this change, ensuring clarity on the behavior of the `weaken_etag_on_compress` option. Signed-off-by: Rafael Antunes <rafael.gaspar@me.com>

rafaelgaspar · 2026-03-24T10:05:23Z

@KBaichoo done, please let me know if there is anything else.

Signed-off-by: Rafael Antunes <rafael.gaspar@me.com>

rafaelgaspar · 2026-03-25T09:23:00Z

@KBaichoo fixed the formatter issue, could you trigger CI again?

…ompressing (envoyproxy#43755) **Commit Message:** compressor: add weaken_etag_on_compress to weaken strong ETags when compressing **Additional Description:** Adds a new option ``weaken_etag_on_compress`` to the compressor filter's ``response_direction_config``. When enabled, strong ``ETag`` response headers are weakened by prepending ``W/`` to the value (e.g. ``"abc123"`` becomes ``W/"abc123"``) instead of being removed when compression is applied. Weak ETags (already starting with ``W/``) are left unchanged. This allows caches and conditional requests to keep working while indicating that the representation was modified by compression, matching behavior used in other proxies (e.g. Varnish). The default remains ``false``, preserving existing behavior (strong ETags are removed when compressing). Fixes envoyproxy#35152 Signed-off-by: Rafael Antunes <rafael.gaspar@me.com> Signed-off-by: Xuyang Tao <taoxuy@google.com>

…ompressing (envoyproxy#43755) **Commit Message:** compressor: add weaken_etag_on_compress to weaken strong ETags when compressing **Additional Description:** Adds a new option ``weaken_etag_on_compress`` to the compressor filter's ``response_direction_config``. When enabled, strong ``ETag`` response headers are weakened by prepending ``W/`` to the value (e.g. ``"abc123"`` becomes ``W/"abc123"``) instead of being removed when compression is applied. Weak ETags (already starting with ``W/``) are left unchanged. This allows caches and conditional requests to keep working while indicating that the representation was modified by compression, matching behavior used in other proxies (e.g. Varnish). The default remains ``false``, preserving existing behavior (strong ETags are removed when compressing). Fixes envoyproxy#35152 Signed-off-by: Rafael Antunes <rafael.gaspar@me.com> Signed-off-by: Jonathan Wu <jtwu@google.com>

…ompressing (envoyproxy#43755) **Commit Message:** compressor: add weaken_etag_on_compress to weaken strong ETags when compressing **Additional Description:** Adds a new option ``weaken_etag_on_compress`` to the compressor filter's ``response_direction_config``. When enabled, strong ``ETag`` response headers are weakened by prepending ``W/`` to the value (e.g. ``"abc123"`` becomes ``W/"abc123"``) instead of being removed when compression is applied. Weak ETags (already starting with ``W/``) are left unchanged. This allows caches and conditional requests to keep working while indicating that the representation was modified by compression, matching behavior used in other proxies (e.g. Varnish). The default remains ``false``, preserving existing behavior (strong ETags are removed when compressing). Fixes envoyproxy#35152 Signed-off-by: Rafael Antunes <rafael.gaspar@me.com> Signed-off-by: Nick Shipilov <nick.shipilov.n@gmail.com>

…ompressing (envoyproxy#43755) **Commit Message:** compressor: add weaken_etag_on_compress to weaken strong ETags when compressing **Additional Description:** Adds a new option ``weaken_etag_on_compress`` to the compressor filter's ``response_direction_config``. When enabled, strong ``ETag`` response headers are weakened by prepending ``W/`` to the value (e.g. ``"abc123"`` becomes ``W/"abc123"``) instead of being removed when compression is applied. Weak ETags (already starting with ``W/``) are left unchanged. This allows caches and conditional requests to keep working while indicating that the representation was modified by compression, matching behavior used in other proxies (e.g. Varnish). The default remains ``false``, preserving existing behavior (strong ETags are removed when compressing). Fixes envoyproxy#35152 Signed-off-by: Rafael Antunes <rafael.gaspar@me.com>

rafaelgaspar requested review from KBaichoo and mattklein123 as code owners March 4, 2026 11:35

rafaelgaspar had a problem deploying to external-contributors March 4, 2026 11:35 — with GitHub Actions Error

repokitteh-read-only Bot added the api label Mar 4, 2026

repokitteh-read-only Bot assigned adisuissa Mar 4, 2026

rafaelgaspar force-pushed the feat/compressor/35152-weaken-etag branch from ba6a487 to 3bcab8f Compare March 4, 2026 11:38

rafaelgaspar had a problem deploying to external-contributors March 4, 2026 11:38 — with GitHub Actions Error

repokitteh-read-only Bot added the waiting label Mar 10, 2026

adisuissa reviewed Mar 10, 2026

View reviewed changes

Comment thread api/envoy/extensions/filters/http/compressor/v3/compressor.proto Outdated

Comment thread api/envoy/extensions/filters/http/compressor/v3/compressor.proto Outdated

rafaelgaspar force-pushed the feat/compressor/35152-weaken-etag branch from 3bcab8f to 974d00d Compare March 10, 2026 22:14

repokitteh-read-only Bot removed the waiting label Mar 10, 2026

rafaelgaspar had a problem deploying to external-contributors March 10, 2026 22:14 — with GitHub Actions Error

rafaelgaspar had a problem deploying to external-contributors March 10, 2026 22:29 — with GitHub Actions Error

rafaelgaspar had a problem deploying to external-contributors March 10, 2026 22:33 — with GitHub Actions Error

adisuissa reviewed Mar 17, 2026

View reviewed changes

repokitteh-read-only Bot removed the api label Mar 17, 2026

repokitteh-read-only Bot assigned KBaichoo Mar 17, 2026

KBaichoo previously approved these changes Mar 19, 2026

View reviewed changes

Comment thread source/extensions/filters/http/compressor/compressor_filter.cc Outdated

Comment thread source/extensions/filters/http/compressor/compressor_filter.cc Outdated

repokitteh-read-only Bot added the waiting label Mar 20, 2026

rafaelgaspar added 3 commits March 23, 2026 10:32

Move new field closer to disable_on_etag_header

4c94362

Signed-off-by: Rafael Antunes <rafael.gaspar@me.com>

rafaelgaspar dismissed KBaichoo’s stale review via 4c94362 March 23, 2026 09:32

rafaelgaspar force-pushed the feat/compressor/35152-weaken-etag branch from 9f2ffb3 to 4c94362 Compare March 23, 2026 09:32

repokitteh-read-only Bot removed the waiting label Mar 23, 2026

rafaelgaspar had a problem deploying to external-contributors March 23, 2026 09:32 — with GitHub Actions Error

rafaelgaspar added 2 commits March 23, 2026 10:40

Refactor ETag handling in CompressorFilter

acaca13

Introduced a helper function `isWeakEtag` to streamline the logic for identifying weak ETags in the `sanitizeEtagHeader` and `weakenEtagHeader` methods. Signed-off-by: Rafael Antunes <rafael.gaspar@me.com>

rafaelgaspar had a problem deploying to external-contributors March 23, 2026 09:41 — with GitHub Actions Error

rafaelgaspar requested a review from KBaichoo March 23, 2026 09:41

KBaichoo reviewed Mar 23, 2026

View reviewed changes

Comment thread source/extensions/filters/http/compressor/compressor_filter.cc Outdated

repokitteh-read-only Bot added the waiting label Mar 23, 2026

repokitteh-read-only Bot removed the waiting label Mar 24, 2026

rafaelgaspar temporarily deployed to external-contributors March 24, 2026 10:04 — with GitHub Actions Inactive

rafaelgaspar requested a review from KBaichoo March 24, 2026 10:04

KBaichoo previously approved these changes Mar 24, 2026

View reviewed changes

Fix linter isssue in CompressorFilter tests

62fea57

Signed-off-by: Rafael Antunes <rafael.gaspar@me.com>

rafaelgaspar dismissed KBaichoo’s stale review via 62fea57 March 25, 2026 09:17

rafaelgaspar temporarily deployed to external-contributors March 25, 2026 16:06 — with GitHub Actions Inactive

KBaichoo approved these changes Mar 25, 2026

View reviewed changes

KBaichoo merged commit 97f4cec into envoyproxy:main Mar 25, 2026
30 checks passed

Conversation

rafaelgaspar commented Mar 4, 2026

Uh oh!

repokitteh-read-only Bot commented Mar 4, 2026

Uh oh!

repokitteh-read-only Bot commented Mar 4, 2026

Uh oh!

agrawroh commented Mar 10, 2026

Uh oh!

agrawroh commented Mar 10, 2026

Uh oh!

adisuissa left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

rafaelgaspar commented Mar 10, 2026

Uh oh!

paul-r-gall commented Mar 16, 2026

Uh oh!

adisuissa left a comment

Choose a reason for hiding this comment

Uh oh!

adisuissa commented Mar 17, 2026

Uh oh!

KBaichoo left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

KBaichoo commented Mar 20, 2026

Uh oh!

rafaelgaspar commented Mar 23, 2026

Uh oh!

KBaichoo left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

rafaelgaspar commented Mar 24, 2026

Uh oh!

rafaelgaspar commented Mar 25, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants