Agent fails build config "Process exited with code 127"

The build fails after entering the first and only step:

Step 1/1: Command Line
14:22:4 Running step within Docker container python:3
14:22:42 Starting: /bin/sh -c ". /opt/buildagent/temp/agentTmp/docker-wrapper-231835785673144342.sh && docker run --rm -w /opt/buildagent/work/3e7a09ce3025fbcf --label jetbrains.teamcity.buildId=524 -v "/opt/buildagent/lib:/opt/buildagent/lib:ro" -v "/opt/buildagent/tools:/opt/buildagent/tools:ro" -v "/opt/buildagent/plugins:/opt/buildagent/plugins:ro" -v "/opt/buildagent/work/3e7a09ce3025fbcf:/opt/buildagent/work/3e7a09ce3025fbcf" -v "/opt/buildagent/temp/agentTmp:/opt/buildagent/temp/agentTmp" -v "/opt/buildagent/temp/buildTmp:/opt/buildagent/temp/buildTmp" -v "/opt/buildagent/system:/opt/buildagent/system" --env-file /opt/buildagent/temp/agentTmp/docker-wrapper-1292889910090267051.envList --entrypoint /bin/sh "python:3" /opt/buildagent/temp/agentTmp/docker-shell-script-5726473492574467882.sh"
14:22:42 in directory: /opt/buildagent/work/3e7a09ce3025fbcf
14:22:42 /bin/sh: 0: Can't open /opt/buildagent/temp/agentTmp/docker-shell-script-5726473492574467882.sh
14:22:43 Process exited with code 127
14:22:43 Process exited with code 127 (Step: Command Line)
14:22:43 Step Command Line failed

The python:3 container, is mounting the agentTmp directory to the same place within itself (-v "/opt/buildagent/temp/agentTmp:/opt/buildagent/temp/agentTmp").

However, after inspecting the agent, we can see the the script is present, so it's not that the script is missing and all users have read permissions:

$ kubectl exec -it teamcity-agent-88466b89d-2phzf /bin/bash
root@teamcity-agent-88466b89d-2phzf:/# cd /opt/buildagent/temp/agentTmp/
root@teamcity-agent-88466b89d-2phzf:/opt/buildagent/temp/agentTmp# ls
build.finish.properties.gz  build.start.properties.gz  docker-shell-script-5726473492574467882.sh  docker-wrapper-1292889910090267051.envList  docker-wrapper-231835785673144342.sh
root@teamcity-agent-88466b89d-2phzf:/opt/buildagent/temp/agentTmp# ls -lah
total 20K
drwxr-xr-x 2 root root  217 Mar 12 14:22 .
drwxr-xr-x 7 root root  118 Mar 12 14:22 ..
-rw-r--r-- 1 root root 3.8K Mar 12 14:22 build.finish.properties.gz
-rw-r--r-- 1 root root 3.8K Mar 12 14:22 build.start.properties.gz
-rw-r--r-- 1 root root  110 Mar 12 14:22 docker-shell-script-5726473492574467882.sh
-rw-r--r-- 1 root root  249 Mar 12 14:22 docker-wrapper-1292889910090267051.envList
-rw-r--r-- 1 root root  788 Mar 12 14:22 docker-wrapper-231835785673144342.sh
root@teamcity-agent-88466b89d-2phzf:/opt/buildagent/temp/agentTmp# 

For the sake of completeness, I also include our Kubernetes deployment config which runs the agent:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: teamcity-agent
spec:
  replicas: 2
  selector:
    matchLabels:
      name: teamcity-agent
  strategy:
    type: Recreate
  template:
    metadata:
      labels:
        name: teamcity-agent
    spec:
      containers:
      - image: jetbrains/teamcity-agent:2019.1-linux
        name: teamcity-agent
        imagePullPolicy: "Always"
        env:
        - name: SERVER_URL
          value: teamcity:8002
        resources:
          requests:
            memory: 1G
          limits:
            memory: 4G
        volumeMounts:
          - name: dockersock
            mountPath: "/var/run/docker.sock"
      volumes:
        - name: dockersock
          hostPath:
            path: "/var/run/docker.sock"

Historically, this has been working fine and this problem has only emerged recently. We're not aware of any change that could have caused this. Any light shed on this would be appreciated.