A docker image for running fluentd in Kubernetes pods.
Based on:
latest
/0.14.14-0.9.9
(Fluentd v0.14.14)
Naming convention for images is $FLUENTD_VERSION
-$KUBE_FLUENTD_VERSION
0.9.9
- Bump Fluentd to 0.14.14 to incorporate the fix for fluent/fluentd#1485
0.9.8
- Set buffer_queue_full_action to
block
for tail plugins so that we won't drop messages anymore
- Set buffer_queue_full_action to
0.9.7
- fluentd.conf is customizable via environment variables (#3, thanks to @cw-hayashi)
0.9.6
- Fix the fluentd OOM issue when ingesting a lot of log
- Also see fluent/fluentd#1260 for why we don't use
overflow_action block
- Also see fluent/fluentd#1260 for why we don't use
- Fix the fluentd OOM issue when ingesting a lot of log
0.9.5
- Remove /var/lib/apt/lists/* to save approximately 20MB of image size
0.9.4
- Build confd outside of containers to save 1MB of image size
0.9.3
- Switch from alpine to ubuntu-slim:0.4
- Use jemalloc
- Fluentd 0.14.9
0.9.2
- I had fixed the wrong part of fluent.conf in 0.9.1. Now I've done it right.
- Bump Fluentd to 0.14.2 to fix the issue reported in the metadata filter repo
0.9.1
- Use multiple threads for processing as per a change in GoogleCloudPlatform/google-fluentd
# (1) Provide GOOGLE_FLEUNTD_* environments appropriate values
# (2) Build the definition for a k8s secret object
$ make fluentd.secret.yaml
# (3) Create the secret object from the definition
$ kubectl create -f fluentd.secret.yaml
# (4) Create a fluentd daemonset that reads and depends on the secret
$ kubectl create -f fluentd.daemonset.yaml