[ https://issues.apache.org/jira/browse/NIFI-5955?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17250872#comment-17250872 ]
Ruben Laguna commented on NIFI-5955: ------------------------------------ The problem with this kind of healthcheck (/nifi) is that the UI may be up (GET /nifi 200 OK) but the UI may still be unusable (for example the UI is a node that is disconnected from the cluster). In [NIFI-8100] I've requested a healthcheck endpoint that signals that the node is actually ready and healthy so that it can be used by external load balancers, but I guess I would be also good to use that new endpoint for the docker healthcheck. > Provide a Docker HEALTHCHECK > ---------------------------- > > Key: NIFI-5955 > URL: https://issues.apache.org/jira/browse/NIFI-5955 > Project: Apache NiFi > Issue Type: Improvement > Components: Docker > Reporter: Anthony Mastrean > Priority: Minor > Time Spent: 4h 20m > Remaining Estimate: 0h > > It takes Nifi a noticeable amount of time to become available at > [http://localhost:8080/nifi.] We can help indicate readiness to both human > and machine users (see > [depends_on|https://docs.docker.com/compose/compose-file/compose-file-v2/#depends_on] > in Compose) by establishing a HEALTHCHECK in the Dockerfile. > The > [documentation|https://docs.docker.com/engine/reference/builder/#healthcheck] > suggests > {code:java} > HEALTHCHECK CMD curl -f http://localhost:8080/nifi || exit 1{code} > > But, there are [better cURL > flags|https://explainshell.com/explain?cmd=curl+-fsSL+example.org] for > scripts. > {code:java} > HEALTHCHECK CMD curl -fsSL http://localhost:8080/nifi || exit 1 > {code} > > And I sometimes see problems using {{localhost}} (possibly a Docker > networking issue?). I've seen this pattern suggested elsewhere. > {code:java} > HEALTHCHECK CMD curl -fsSL http://$(hostname -i || echo localhost):8080/nifi > || exit 1 > {code} > -- This message was sent by Atlassian Jira (v8.3.4#803005)