[
https://issues.apache.org/jira/browse/FLINK-15448?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17009825#comment-17009825
]
Till Rohrmann commented on FLINK-15448:
---------------------------------------
Thanks for starting this discussion [~victor-wong]. I think you brought up a
quite important point which is that Flink uses a lot of ids in its logs which
are hard to decipher for the user. One example are the {{ResourceIDs}} and
{{InstanceIDs}} of the {{TaskExecutors}}. Another example is the usage of
{{SlotRequestId}} for various logical slots and {{AllocationID}}. I believe
Flink's logs can be cleaned up big time by associating more meaning with ids.
Often it is a good idea to compose ids from higher level ids in order to attach
meaning to them. So for example, the {{InstanceID}} could be composed of the
{{ResourceID}} plus a monotonically increasing value. Of course, the initial
part of the id should be descriptive as well. Here I agree with [~victor-wong]
that it would be super helpful if for the {{ResourceID}} it could be something
to identify the machine on which it is running. Of course, this might cause
that we log an information redundantly in the logs.
I see the point that [~xintongsong] raised with breaking existing solutions if
we change the ids. This is indeed not optimal but I would be willing to do it
in order to improve the readability of the logs for our users. Technically, we
never gave guarantees for the logs.
If we touch the {{ResourceID}} now, then we should also have a plan for the
other distributed components which use the {{ResourceID}} to identify
themselves.
> Log host informations for TaskManager failures.
> -----------------------------------------------
>
> Key: FLINK-15448
> URL: https://issues.apache.org/jira/browse/FLINK-15448
> Project: Flink
> Issue Type: Improvement
> Components: Runtime / Coordination
> Affects Versions: 1.9.1
> Reporter: Victor Wong
> Assignee: Victor Wong
> Priority: Minor
>
> With Flink on Yarn, sometimes we ran into an exception like this:
> {code:java}
> java.util.concurrent.TimeoutException: The heartbeat of TaskManager with id
> container_xxxx timed out.
> {code}
> We'd like to find out the host of the lost TaskManager to log into it for
> more details, we have to check the previous logs for the host information,
> which is a little time-consuming.
> Maybe we can add more descriptive information to ResourceID of Yarn
> containers, e.g. "container_xxx@host_name:port_number".
> Here's the demo:
> {code:java}
> class ResourceID {
> final String resourceId;
> final String details;
> public ResourceID(String resourceId) {
> this.resourceId = resourceId;
> this.details = resourceId;
> }
> public ResourceID(String resourceId, String details) {
> this.resourceId = resourceId;
> this.details = details;
> }
> public String toString() {
> return details;
> }
> }
> // in flink-yarn
> private void startTaskExecutorInContainer(Container container) {
> final String containerIdStr = container.getId().toString();
> final String containerDetail = container.getId() + "@" +
> container.getNodeId();
> final ResourceID resourceId = new ResourceID(containerIdStr,
> containerDetail);
> ...
> }
> {code}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
