[ https://issues.apache.org/jira/browse/HADOOP-7973?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Daryn Sharp updated HADOOP-7973: -------------------------------- Attachment: HADOOP-7973-3.patch Idea is basically to allow MR to be backward-compatible by maintaining cached FS objects that are distinct from the FS objects used by user code. That's exactly how it used to work due to the DFS port-stripping bug. I pulled in minimal changes from trunk to allow a unique id in the FS cache key. I didn't pull in the new APIs, but rather smuggled a unique id in via a config setting. It's not the cleanest, but I'm open to alternatives. I've probably missed a few spots in the MR framework, but I'd like comments on the approach before I go further. > DistributedFileSystem close has severe consequences > --------------------------------------------------- > > Key: HADOOP-7973 > URL: https://issues.apache.org/jira/browse/HADOOP-7973 > Project: Hadoop Common > Issue Type: Bug > Components: fs > Affects Versions: 1.0.0 > Reporter: Daryn Sharp > Assignee: Daryn Sharp > Priority: Blocker > Attachments: HADOOP-7973-2.patch, HADOOP-7973-3.patch, > HADOOP-7973.patch > > > The way {{FileSystem#close}} works is very problematic. Since the > {{FileSystems}} are cached, any {{close}} by any caller will cause problems > for every other reference to it. Will add more detail in the comments. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira