[JENKINS-25369] DNS multicast error messages

[timja]

I've seen these error messages in my console output, which comes from jmdns.

The level of the logging messages should be toned down significantly given that these are from a relatively minor feature in Jenkins.

---

Originally reported by uday, imported from: DNS multicast error messages

• assignee: fcojfernandez
• status: Resolved
• priority: Minor
• resolution: Fixed
• resolved: 2019-05-21T15:09:49+00:00
• imported: 2022/01/10

[timja]

danielbeck:

Since they're in a third party component, I don't think effort/gain works out in favor of reducing log level.

Workaround: java -Dhudson.DNSMultiCast.disabled=true

[timja]

derfunk:

unintentional self assign, sorry.

[timja]

ayursha:

So is it going to be fixed as this logging spam is pretty annoying? jmnds original package has not been updated for a while, so we could either fix the fork or drop this dependency completely if we don't need that functionality.

[timja]

danielbeck:

It doesn't look to me like it's that annoying if you're not even going to disable the feature and restart Jenkins once in 6+ weeks.

[timja]

neerajlad:

Does this could harm any execution? since i ma also getting same this when i run sonar-plugin into jenkins.

[timja]

kbaltrinic:

For those with size based log rotation strategies, this is more than an annoyance; it greatly diminishes the amount of meaningful log data that gets retained. Sampling my current logs, this message looks to comprise 70 - 80% of Jenkins log output. I consider that to be significant.

[timja]

danielbeck:

kbaltrinic The feature causing the log messages can easily be disabled from the next time you start Jenkins again, see my comment from 30/Oct/14 1:31 AM. There's a good chance you're not even using it.

[timja]

dzschille:

This error message flooded our log this weekend, till there was no disc space left. The log file was more than 430 GB big. The messages mostly like this one: https://gist.github.com/dzschille/ed739c040b0a5ba4d23c

[timja]

danielbeck:

Related to JENKINS-10160, possibly a duplicate.

[timja]

oleg_nenashev:

It may happen even without network issues, e.g. when there is an unsupported DNS Message type coming from DNS. We use an old library version, and not all DNS Message types are supported (discovered in JENKINS-50812)

Record type enumeration in the lib: https://github.com/jenkinsci/jmdns/blob/64f48ab717e0c8022f60d7dcb7f148f47c7f1227/src/main/java/javax/jmdns/impl/constants/DNSRecordType.java

List of all DNS record types: https://en.wikipedia.org/wiki/List_of_DNS_record_types . Likely some types have been added in new DNS versions via RFCs. E.g. URI identifier has been added in 2015 by https://tools.ietf.org/html/rfc7553 , so no wonder it's not supported by the lib.

There is https://github.com/jenkinsci/jmdns/pull/2 , which may resolve the issue by bumping to the new version

[timja]

fcojfernandez:

I've been doing some research here and updating the library seems to fix the problem. At least is a starting point. 

As I can see, the library was forked just because we want to use an abort method instead of the close method provided by jmdns. Moving the logic of this abort method to the core, probably we can put aside the fork and use the library instead, so any update would be easier. Opening a PR soon.

[timja]

oleg_nenashev:

It was released in Jenkins 2.178, but we do not think backporting is justified for now

[timja]

oleg_nenashev:

LTS backporting comment: let's do more testing before considering backporting

[timja]

olivergondza:

oleg_nenashev, fcojfernandez, have you gain enough confidence in backporting https://github.com/jenkinsci/jenkins/pull/4021 and https://github.com/jenkinsci/jenkins/pull/4037 into .2?

[timja]

olivergondza:

I am rejecting the backport to be added to 2.176 LTS line. Will appear in the next one naturally.

[timja]

[Originally duplicated by: JENKINS-10160]

[timja]

[Originally duplicated by: JENKINS-29490]

[timja]

[Originally duplicated by: JENKINS-50812]