Newest version of the container does not start up due to config errors

[WolfgangMehner]

We are running a fluentd Daemonset on our Kubernetes cluster (v.1.19.6 on EKS)

We started up the most recent version of the image yesterday, and everything worked fine. We used the tag with the resulting digest:
v1-debian-elasticsearch7 -> 18d5a7158b5b
Today we started up containers with the same config and tag, but got a different container that did not start up anymore:
v1-debian-elasticsearch7 -> 2a7e2d1e42f3

When the container started up, we got an error along the lines of:

ConfigError error="<parse> section is required."

Unfortunately we can not provide the precise error message anymore, since nothing was logged with the fluentd PODs not starting up.

[moorepatrick]

I'm seeing the same thing. A minor version was pushed last night but you were probably tagged on the major version.
v1-debian-elasticsearch7 is aliasing v1.12.0-debian-elasticsearch7-1.1 today, so try v1.12.0-debian-elasticsearch7-1.0
(Note that you need the full v1.12.0-debian)

In my case I'm seeing the addition of gem 'fluent-plugin-parser-cri' version '0.1.0' at startup, which looks to be a likely culprit. Still digging into a real solution.

Related: #521