Mention the path of DLQ to indicate DLQ if full for which pipeline

[amitavmohanty01]

A series of messages of the following form do not indicate DLQ for which pipeline is full.
[2019-10-30T11:53:33,801][ERROR][org.logstash.common.io.DeadLetterQueueWriter] cannot write event to DLQ: reached maxQueueSize of 1073741824

So, it is probably better to let the user know the path of the DLQ file so that the user can:

1. know which pipeline is affected
2. clear the DLQ files manually if appropriate

[elasticmachine]

Since this is a community submitted pull request, a Jenkins build has not been kicked off automatically. Can an Elastic organization member please verify the contents of this patch and then kick off a build manually?

[amitavmohanty01]

@danhermann please provide feedback on the approach and implementation

[danhermann]

@amitavmohanty01, this looks pretty good to me. There are a couple code formatting issues, but those are pretty minor. The line above should include spaces in the places where I inserted underscores below:

logger.error("cannot write event to DLQ_(path:_"_+ this.queuePath +_"): reached maxQueueSize of " + maxQueueSize);

I'll need to defer to someone on the @elastic/logstash team to approve and merge this PR, though.

[amitavmohanty01]

I have modified as suggested and edited the commit.

[andsel]

@amitavmohanty01 with PR #11075 there should already be the name of failing pipeline in the logs which cover point 1).
This PR is good for the point 2) where the user could now where is the directory to cleanup

[amitavmohanty01]

@andsel thanks for the inputs. I am in agreement.

Can you please trigger a Jenkins build so that we can be sure there is nothing missed out?

[andsel]

Jenkins test this please

[andsel]

@amitavmohanty01 The error should be resolved if you align the code to master (latest commits solved it):

$ git checkout master
$ git fetch upstream
$ git merge upstream/master
$ git checkout dlq_file_path
$ git rebase master
$ git push origin dlq_file_path -f

[amitavmohanty01]

@andsel I have rebased on master. Please trigger the Jenkins build again.

[amitavmohanty01]

@andsel ping

[andsel]

Jenkins test this please

[amitavmohanty01]

@andsel it seems the errors on Jenkins are unrelated to my code. Am I missing something?

[andsel]

@amitavmohanty01 to me seems all ok, I'll asking Jenkins to rebuild it. If it fails again probably there is a subtle dependency between the two facts to spoil

[andsel]

Jenkins test this please

[amitavmohanty01]

@andsel I ran ./gradlew test locally and the result was the following.

----------------------------------------------------------------------
|  Results: SUCCESS (56 tests, 56 successes, 0 failures, 0 skipped)  |
----------------------------------------------------------------------

Is the Jenkins environment having any nuances causing the failures? I am considering that especially because the failure and the code change seem unrelated.

[andsel]

@amitavmohanty01 to build LogStash you have to execute ./gradlew build from the checkout directory. But the Gradle task that fails is :logstash-xpack:rubyIntegrationTests. It's not related to your PR but something to Elasticsearch master (elastic/elasticsearch#48170)

[andsel]

@amitavmohanty01 with the integration of PR #11297 the issue is fixed. Please could you rebase your branch to the current master?

[amitavmohanty01]

@andsel done

[andsel]

Jenkins test this please

[andsel]

Jenkins test this please

[andsel]

Jenkins test this please

[elasticsearch-bot]

Andrea Selva merged this into the following branches!

Branch	Commits
master	b99c2f9
7.x	3dd117a

[colinsurprenant]

Adding a note here to not forget to backport on 7.5.0 once released so that it gets into 7.5.1 since we are late in the 7.5.0 release process.

[colinsurprenant]

@andsel also, we should backport to 7.4 branch!

[elasticsearch-bot]

Andrea Selva merged this into the following branches!

Branch	Commits
7.4	d65ef8e

[elasticsearch-bot]

João Duarte merged this into the following branches!

Branch	Commits
7.5	b50ff4f