Kibana version: 6.7-8.0/master
Describe the bug:
state: running foreverlog [16:33:44.786] [warning][task_manager] The task maps_telemetry "Maps-maps_telemetry" is not cancellable.Steps to reproduce:
async () => {
return new Promise(() => {
console.log('waiting for nothing to happen');
});
}
waiting for nothing to happen
server log [20:11:13.361] [warning][task_manager] The task demo_form_scheduler "9--uzmgBXf7pSraMfOwb" is not cancellable.
waiting for nothing to happen
server log [20:16:13.984] [warning][task_manager] The task demo_form_scheduler "9--uzmgBXf7pSraMfOwb" is not cancellable.
waiting for nothing to happen
server log [20:21:14.973] [warning][task_manager] The task demo_form_scheduler "9--uzmgBXf7pSraMfOwb" is not cancellable.
waiting for nothing to happen
server log [20:26:15.838] [warning][task_manager] The task demo_form_scheduler "9--uzmgBXf7pSraMfOwb" is not cancellable.
waiting for nothing to happen
Expected behavior:
Task Manager should mark the task as failed, remove the task instance, but ultimately stop re-trying
Pinging @elastic/kibana-platform
I ran into this too at some point. not sure how to reproduce.
@tsullivan @aaronjcaldwell to remove the warning clutter from the console coming from the maps-app, would it be just sufficient to implement a cancel method on the maps-telemery task?
@thomasneirynck it's definitely a Task Manager / Platform bug. If the task runner doesn't return a cancellable promise, you should see a warning log 3 times (given retry logic is happening). But then the task should be marked as failed and not retried.
How do you solve the problem, if the task never completes?
It would be nice if the Task Manager exposed a cancelable promise utility that plugins can use as a helper. I haven't worked on cancelable promises before, but there are/have been examples in other parts of the Kibana code that use Rxjs.
@chrisdavies had an idea to expose the helper in Task Manager phase 1, but it got scrapped.
How do you solve the problem, if the task never completes?
If the task never completes, the timeout check in Task Manager should mark it as a failed run. It will re-try after a backoff period. If the task fails X number of times, Task Manager will unschedule the task. Through this, there should be useful logging - a task that isn't completing is a bug that should be fixed.
Hi everyone!
I installed ELK from repo v7 on the Fedora30 server.
And i have similar problem.
Can you help me?
I'm having the same problem with elasticsearch and kibana 7.2.0 in k8s.
Pinging @elastic/kibana-stack-services
ubuntu 16.04 with kibana 7.2.0 - also have an issue
However after downgraded to 7.1.1 it is working properly.
same problem on, CentOS kibana-7.3.0-1.x86_64
The message started appearing after updating from 7.2.0 to 7.3.0.
It repeats every 1 minute.
Edit: the error message was displayed when Kibana + Elasticsearch was updated but only Kibana was automatically restarted. Restarting Elasticsearch made the warning (and the obvious "Kibana requires Elasticsearch 7.3.0"...) go away.
Pinging @elastic/kibana-gis
Can someone on the maps team look at implementing a cancel function for the maps telemetry task?
My understanding was that this was an issue upstream in telemetry that happens to manifest in the maps app. I'll take a fresh look though and see if I can determine the exact issue
Task manager expects that tasks have cancel functions, and the maps telemetry one does not. There may also be an upstream issue, but the lack of a cancel function is definitely itself an issue.
Most helpful comment
I'm having the same problem with elasticsearch and kibana
7.2.0in k8s.