airflow: Scheduler livenessProbe errors on new helm chart

Official Helm Chart version

1.6.0 (latest released)

Apache Airflow version

v2.1.2

Kubernetes Version

v1.22.10 (GKE version v1.22.10-gke.600)

Helm Chart configuration

Only livenessProbe config before and during the issue:

# Airflow scheduler settings
scheduler:
  livenessProbe:
    initialDelaySeconds: 10
    timeoutSeconds: 15
    failureThreshold: 10
    periodSeconds: 60

Docker Image customisations

Here’s the image we use based on the apache airflow image:

### Main official airflow image
FROM apache/airflow:2.1.2-python3.8

USER root

RUN apt update

RUN echo 'debconf debconf/frontend select Noninteractive' | debconf-set-selections

### Add OS packages here
## GCC compiler in case it's needed for installing python packages
RUN apt install -y -q build-essential

USER airflow

### Changing the default SSL / TLS mode for mysql client to work properly
## https://askubuntu.com/questions/1233186/ubuntu-20-04-how-to-set-lower-ssl-security-level
## https://bugs.launchpad.net/ubuntu/+source/mysql-8.0/+bug/1872541
## https://stackoverflow.com/questions/61649764/mysql-error-2026-ssl-connection-error-ubuntu-20-04
RUN echo $'openssl_conf = default_conf\n\
[default_conf]\n\
ssl_conf = ssl_sect\n\
[ssl_sect]\n\
system_default = ssl_default_sect\n\
[ssl_default_sect]\n\
MinProtocol = TLSv1\n\
CipherString = DEFAULT:@SECLEVEL=1' >> /home/airflow/.openssl.cnf
## OS env var to point to the new openssl.cnf file
ENV OPENSSL_CONF=/home/airflow/.openssl.cnf

### Add airflow providers
RUN pip install apache-airflow-providers-apache-beam
### End airflow providers

### Add extra python packages
RUN pip install python-slugify==3.0.3
### End extra python packages

What happened

After the upgrade to the helm chart 1.6.0, the scheduler POD was restarting as the livenessProbe was failing.

Command for the new livenessProbe from helm chart 1.6.0 tested directly on our scheduler POD:

airflow@yc-data-airflow-scheduler-0:/opt/airflow$ CONNECTION_CHECK_MAX_COUNT=0 AIRFLOW__LOGGING__LOGGING_LEVEL=ERROR exec /entrypoint \
> airflow jobs check --job-type SchedulerJob --hostname $(hostname)
No alive jobs found.

Removing the --hostname argument works and a live job is found:

airflow@yc-data-airflow-scheduler-0:/opt/airflow$ CONNECTION_CHECK_MAX_COUNT=0 AIRFLOW__LOGGING__LOGGING_LEVEL=ERROR exec /entrypoint \
> airflow jobs check --job-type SchedulerJob
Found one alive job.

What you think should happen instead

livenessProbe should not error.

How to reproduce

No response

Anything else

No response

Are you willing to submit PR?

• Yes I am willing to submit a PR!

Code of Conduct

• I agree to follow this project’s Code of Conduct