[LIVY-878] Replace log4j with reload4j - #392
Conversation
ksumit
left a comment
There was a problem hiding this comment.
I noticed following in mvn dependency:tree output, the scope is test, so not that worrisome, but thought to bring to your attention.
[INFO] +- org.apache.directory.server:apacheds-server-integ:jar:2.0.0.AM26:test
[INFO] | +- org.apache.directory.server:apacheds-core-annotations:jar:2.0.0.AM26:test
[INFO] | | +- org.apache.directory.mavibot:mavibot:jar:1.0.0-M8:test
[INFO] | | \- org.apache.directory.server:apacheds-mavibot-partition:jar:2.0.0.AM26:test
[INFO] | +- org.apache.directory.server:apacheds-core-jndi:jar:2.0.0.AM26:test
[INFO] | +- org.apache.directory.server:apacheds-interceptor-kerberos:jar:2.0.0.AM26:test
[INFO] | +- org.apache.directory.server:apacheds-server-annotations:jar:2.0.0.AM26:test
[INFO] | | \- org.apache.directory.server:apacheds-protocol-kerberos:jar:2.0.0.AM26:test
[INFO] | +- org.apache.directory.server:apacheds-test-framework:jar:2.0.0.AM26:test
[INFO] | | \- org.apache.directory.api:api-ldap-codec-standalone:jar:2.0.0:test
[INFO] | | \- org.apache.directory.api:api-ldap-net-mina:jar:2.0.0:test
[INFO] | +- ldapsdk:ldapsdk:jar:4.1:test
[INFO] | +- log4j:log4j:jar:1.2.17:test
|
If it isn't getting used, its better if we could exclude the log4j dependency completely even from test scope |
|
AhI thought I addressed the @ayushtkn Thanks, latest push should be better. |
|
Will trigger CI with a push today. |
7ff3ec1 to
a59259c
Compare
Codecov Report
@@ Coverage Diff @@
## master #392 +/- ##
============================================
+ Coverage 65.95% 68.49% +2.53%
- Complexity 837 859 +22
============================================
Files 103 103
Lines 6063 6063
Branches 913 913
============================================
+ Hits 3999 4153 +154
+ Misses 1539 1344 -195
- Partials 525 566 +41 see 25 files with indirect coverage changes 📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more |
|
Bummer, CI failed. Will take a look either later today or Monday. |
|
Integration tests are working, but unit tests still failing with, what I think is an error related to slf4j/log4j loading. |
|
Fixed the Unit tests are failing, however I think it's due to a flaky test based on behavior observed in CI where unit tests pass on the PR but fail once merged into master and vice-cersa. There's an old mailing list thread that also discusses the issue, but with no resolution. |
|
Interesting, this test does seem to be flaky. The recent post-commit tests have succeeded, but that may be just pure luck. Could you please retrigger the tests on this one? If this test proves to be flaky, we should file a separate JIRA for it. Also, thanks for finding that old email thread about this test! |
|
I've been poking at the failure for this change more and turns out it's something different. When I add exclusions for
The I'm still trying to figure out what exactly is going on there but not having much luck. In most other PRs I've seen for reload4j, |
|
Might have figured this out! Swapping |
|
@dacort - looks like the same failures? |
|
@lmccay Yes but I've got another working branch over here https://github.com/apache/incubator-livy/tree/fix/log4j-reload4j-part2 that appears to be working. Just need to verify everything looks good. |
|
@dacort - curious where you are with the separate branch and plan here? |
9a0f832 to
c616540
Compare
I'll check, but it is defined as a non-provided dependency in the main |
|
Given that this has taken us quiet some time, i do want to checkpoint this work and iterate over if there are issues. Approving for check-in. +1, LGTM |
* Replace log4j with reload4j along with relevant exclusions in dependencies * Bump surefire to fix build * Swap slf4j-api with slf4j-reload4j
What changes were proposed in this pull request?
spark-core,hadoop-common(and others),zookeeper, andapacheds-server.How was this patch tested?