From 12bd51fd547b67aa46c899e147027b5644912109 Mon Sep 17 00:00:00 2001
From: Frank Hamand <frankhamand@gmail.com>
Date: Mon, 12 Aug 2024 10:06:32 +0100
Subject: [PATCH] feat: use patched vector to fix 400 responses and to add
 response body feature (#24302)

---
 .github/workflows/vector-docker-build-deploy.yml |  1 +
 vector/replay-capture/Dockerfile                 |  5 ++++-
 vector/replay-capture/tests.yaml                 |  2 ++
 vector/replay-capture/vector.yaml                | 14 ++++++++++++--
 4 files changed, 19 insertions(+), 3 deletions(-)

diff --git a/.github/workflows/vector-docker-build-deploy.yml b/.github/workflows/vector-docker-build-deploy.yml
index 44b86c24cd22a..940b87655a462 100644
--- a/.github/workflows/vector-docker-build-deploy.yml
+++ b/.github/workflows/vector-docker-build-deploy.yml
@@ -76,6 +76,7 @@ jobs:
     deploy:
         runs-on: ubuntu-latest
         needs: build
+        if: github.ref == 'refs/heads/master'
         steps:
             - name: get deployer token
               id: deployer
diff --git a/vector/replay-capture/Dockerfile b/vector/replay-capture/Dockerfile
index e32ab326aadd7..0715ef32a54ae 100644
--- a/vector/replay-capture/Dockerfile
+++ b/vector/replay-capture/Dockerfile
@@ -7,6 +7,9 @@ COPY vector.yaml .
 # evaluate with yq, basically to expand anchors (which vector doesn't support)
 RUN yq -i e 'explode(.)' vector.yaml
 
-FROM timberio/vector:0.40.X-alpine
+# fork of vector from this branch: https://github.com/frankh/vector/tree/main
+# includes 2 fixed: 1 to make Kafka temporary errors return 500 not 400
+# and 1 to allow us to set response body on the http source
+FROM frankh/vector:0.41.0-patched
 
 COPY --from=config-builder /config/vector.yaml /etc/vector/vector.yaml
diff --git a/vector/replay-capture/tests.yaml b/vector/replay-capture/tests.yaml
index 1c30b243c480c..d36ea0ea88019 100644
--- a/vector/replay-capture/tests.yaml
+++ b/vector/replay-capture/tests.yaml
@@ -44,6 +44,8 @@ tests:
                 ."_" = "123456789"
                 %token = "limited_token"
 
+                # we can't properly check things that we verify in the capture decoding vrl sadly :(
+                %quota_limited = true
       outputs:
           - conditions:
                 - source: |
diff --git a/vector/replay-capture/vector.yaml b/vector/replay-capture/vector.yaml
index 8bcc51a10676f..feb30ef7fa3cc 100644
--- a/vector/replay-capture/vector.yaml
+++ b/vector/replay-capture/vector.yaml
@@ -32,6 +32,7 @@ sources:
         query_parameters:
             - _
         host_key: ip
+        response_body_key: '%response'
         decoding:
             codec: vrl
             vrl:
@@ -62,6 +63,16 @@ sources:
                     assert!(is_string(.message[0].properties."$$session_id"), "$$session_id is required")
                     assert!(is_string(%token), "token is required")
 
+                    _, err = get_enrichment_table_record("quota_limited_teams", { "token": %token })
+                    # get_enrichment_table_record returns an err if record is not found
+                    # that means err == null iff we found the quota limited record
+                    %quota_limited = err == null
+
+                    %response = {"status": 1}
+                    if %quota_limited {
+                      %response.quota_limited = ["recordings"]
+                    }
+
 transforms:
     quota_check:
         type: route
@@ -71,8 +82,7 @@ transforms:
             quota_limited:
                 type: vrl
                 source: |
-                    _, err = get_enrichment_table_record("quota_limited_teams", { "token": %token })
-                    err == null  # err is not null if row not found, we want to drop where the row _is_ found
+                    %quota_limited == true
 
     events_parsed:
         type: remap