Search code examples
loggingweblogicweblogic-10.x

Log rotation error in Weblogic server


I am using Weblogic 10.3.4. I am facing an error when starting my server. I have not changed any server start configuration since the last time it started successfully.

This is the error I am getting in nodemanager.log file.

java.io.IOException: Could not rotate server output log file (rename from 'D:\GBKProject\logs\GBK120ServerOut.log' to 'D:\GBKProject\logs\GBK120ServerOut.log00001' failed).
    at weblogic.nodemanager.server.LogFileRotationUtil.rotateLogFile(LogFileRotationUtil.java:75)
    at weblogic.nodemanager.server.LogFileRotationUtil.rotateServerFiles(LogFileRotationUtil.java:38)
    at weblogic.nodemanager.server.AbstractServerManager.startServer(AbstractServerManager.java:203)
    at weblogic.nodemanager.server.AbstractServerManager.start(AbstractServerManager.java:181)
    at weblogic.nodemanager.server.ServerManager.start(ServerManager.java:23)
    at weblogic.nodemanager.server.Handler.handleStart(Handler.java:609)
    at weblogic.nodemanager.server.Handler.handleCommand(Handler.java:121)
    at weblogic.nodemanager.server.Handler.run(Handler.java:71)
    at java.lang.Thread.run(Thread.java:744)

I tried creating an empty file GBK120ServerOut.log00001 in the path, still didn't work. Anyone help me out why this error?


Solution

  • Nevermind, I found the answer from Oracle Support. Anyone having access, see this link.

    https://support.oracle.com/epmos/faces/DocumentDisplay?_afrLoop=537538266169566&id=1261574.1&_afrWindowMode=0&_adf.ctrl-state=8bg20uh8m_4

    Here's their clarification.

    Problem Statement : Several Domains on the same box, using NodeManager, is intermittently having fatal exception when trying to restart/stop one server.

    Issue Clarification: Customer has many domains with multiple vm's on a single physical box. Often when they stop/start only 1 of those, they will not be able to restart it because the .out file will be referenced by a file handle in another java process in the same domain, as well as either none, one or all nodemanagers installed on that box.

    Oracle has a patch available for this bug in aforementioned link, but it needs access privilege. I restarted my system and this locking issue was solved.