Details
-
Bug
-
Resolution: Won't Fix
-
Low
-
None
-
3.2.1
-
None
Description
After upgrading to 3.2.1, I started noticing the following exception in the log:
java.lang.NullPointerException
at com.atlassian.plugin.webresource.PluginResourceLocatorImpl.locateBatchPluginResource(PluginResourceLocatorImpl.java:148)
at com.atlassian.plugin.webresource.PluginResourceLocatorImpl.getDownloadableResource(PluginResourceLocatorImpl.java:85)
at com.atlassian.plugin.servlet.PluginResourceDownload.serveFile(PluginResourceDownload.java:54)
at com.atlassian.plugin.servlet.AbstractFileServerServlet.doGet(AbstractFileServerServlet.java:32)
at javax.servlet.http.HttpServlet.service(HttpServlet.java:796)
at javax.servlet.http.HttpServlet.service(HttpServlet.java:917)
at org.apache.catalina.core.ApplicationFilterChain.servletService(ApplicationFilterChain.java:398)
at org.apache.catalina.core.ApplicationDispatcher.invoke(ApplicationDispatcher.java:792)
at org.apache.catalina.core.ApplicationDispatcher.doForward(ApplicationDispatcher.java:472)
at org.apache.catalina.core.ApplicationDispatcher.forward(ApplicationDispatcher.java:353)
at org.tuckey.web.filters.urlrewrite.RewrittenUrl.doRewrite(RewrittenUrl.java:176)
at org.tuckey.web.filters.urlrewrite.UrlRewriteFilter.doFilter(UrlRewriteFilter.java:728)
at org.apache.catalina.core.ApplicationFilterChain.internalDoFilter(ApplicationFilterChain.java:217)
at org.apache.catalina.core.ApplicationFilterChain.doFilter(ApplicationFilterChain.java:185)
at com.atlassian.gzipfilter.GzipFilter.doFilterInternal(GzipFilter.java:75)
at com.atlassian.gzipfilter.GzipFilter.doFilter(GzipFilter.java:51)
several different paths are affected, for example:
- /s/1636/4/1.0.2/_/download/batch/com.atlassian.auiplugin:jquery-base/com.atlassian.auiplugin:jquery-base.js
- /s/1636/4/1.0.2/_/download/batch/com.atlassian.auiplugin:dialog/com.atlassian.auiplugin:dialog.css
- /s/1636/4/1.0.2/_/download/batch/com.atlassian.auiplugin:drop-down/com.atlassian.auiplugin:drop-down.css
Upon further debugging I noticed that the requests where coming mostly from google webcache that points to assets on the live site. So it looks like the superbatch introduced in 3.2, replaced lots of individual assets that however are still being requested by clients with stale cache. This is mostly harmless, except that our production logs are being polluted with tons of error messages and stack traces that make it difficult to see other problems. This should be an easy, but valuable fix.
