Re: New geode-log4j module

2019-09-25 Thread Anthony Baker
Great writeup, thanks for sharing Kirk! Anthony > On Sep 24, 2019, at 10:24 AM, Kirk Lund wrote: > > All classes that use *log4j-core* have now moved to the new module > *geode-log4j > *on develop. The default log4j2.xml configuration file for Geode Locators > and Servers has also moved to

Re: New geode-log4j module

2019-09-24 Thread Kirk Lund
Please review https://github.com/apache/geode/pull/4089 for the fix. Thanks, Kirk On Tue, Sep 24, 2019 at 1:17 PM Kirk Lund wrote: > It's a race condition in how GfshRule, GfshExecution, and ProcessLogger > fork a process before attaching the stdout and stderr listeners. The > changes for

Re: New geode-log4j module

2019-09-24 Thread Kirk Lund
It's a race condition in how GfshRule, GfshExecution, and ProcessLogger fork a process before attaching the stdout and stderr listeners. The changes for geode-log4j caused this race condition window to take slightly longer. I'll have a fix soon and ConnectCommandAcceptanceTest should then be back

Re: New geode-log4j module

2019-09-24 Thread Kirk Lund
It's not a problem in TomcatInstall. Nothing obvious anywhere. Even the output that GfshRule says is missing the string actually does contain the missing text. I have no idea why this passed in precheckin but fails now... On Tue, Sep 24, 2019 at 11:16 AM Kirk Lund wrote: >

Re: New geode-log4j module

2019-09-24 Thread Kirk Lund
ConnectCommandAcceptanceTest is now failing on develop. Looks like the test is using Geode 1.3.0 (not sure why). It also looks like my commit and the following commit didn't merge properly despite not having any conflicts. Looking into how to fix it now. It's possible that GEODE-7168 removed