CASSJAVA-128 Upgrade Netty to 4.1.130 on 3.x - #2089
Conversation
| <netty.version>4.1.119.Final</netty.version> | ||
| <netty-tcnative.version>2.0.70.Final</netty-tcnative.version> | ||
| <netty.version>4.1.130.Final</netty.version> | ||
| <netty-tcnative.version>2.0.74.Final</netty-tcnative.version> |
There was a problem hiding this comment.
Following convention tcnative is updated to the version specified in the corresponding Netty POM
|
@tolbertam mentioned that we should include the lz4 updates that were on the original 4.x PR (since that was the primary point of the PR in question). Only problem is the version of the lz4 impl we switched to in that PR is built for Java7 and up... and we need that version to get the CVE fix. Still discussing what to do about this. |
… OSGi test issues. Also remove jackson-databind.version var (since it just duplicates jackson.version var)
Would be nice if we could do away with the log4j stuff here but I think that's a bigger lift.
….x to 3.x" This reverts commit f156a05.
…is port to be available based on seed provider configs so changing it is a bit of an issue.
…4.x to 3.x" This reverts commit fca710c.
…moval of storage port randomization
| <url.version>2.4.0</url.version> | ||
| <felix.version>7.0.1</felix.version> | ||
| <pax-exam.version>4.13.4</pax-exam.version> | ||
| <url.version>2.6.4</url.version> |
There was a problem hiding this comment.
The versions currently in use on trunk
| <metrics.version>3.2.2</metrics.version> | ||
| <snappy.version>1.1.2.6</snappy.version> | ||
| <lz4.version>1.4.1</lz4.version> | ||
| <lz4.version>1.10.3</lz4.version> |
There was a problem hiding this comment.
Has to be at least 1.10.2 in order to benefit from this fix. @lukasz-antoniak has already identified this issue in his work on trunk but somehow I missed his change so I kept trying to get 1.10.1 to work. It didn't (and it can't). Fighting with Maven. Extended conversations with Gemini. Lots of dead ends at the end of blind alleys.
This one took a few years off my life.
Side note: Gemini really did drop the ball here. It spent a great deal of time talking to me about Maven's rules for resolving properties in submodules without ever finding the issue in question (or Lukasz's prior work).
| if (cassandraVersion.compareTo(VersionNumber.V4_1_0) < 0) return originalValueStr; | ||
| if (originalKey.endsWith(IN_MS_STR)) return originalValueStr + "ms"; | ||
| if (originalKey.endsWith(IN_KB_STR)) return originalValueStr + "KiB"; | ||
| return originalValueStr; |
There was a problem hiding this comment.
Cassandra 4.1/5.0 aware logic copied directly from trunk
|
Changes at this point build and test almost entirely pass. We've got a few stragglers that appear to be flaky tests but we can fix those after the fact; the larger concern is to get us bumped to Java8 so we can move forward with a 3.13.0. |
2 participants
Mirroring similar work done for 4.x