steveloughran commented on issue #624: HADOOP-15999. S3Guard: Better support for out-of-band operations URL: https://github.com/apache/hadoop/pull/624#issuecomment-475037885 Given I've seen the failure in a standalone run, it's not FS caching somehow meaning raw has picked up a DDB table. Though I've added a way to query S3A FS instance for its auth status to make sure we know what's going on in inside Possible causes then 1. writes are so close together they share the same timestamp 1. the `rawFS.getFileStatus()` call after a read *is picking up the out of date value* 1. the check for modtime is a bit brittle, especially given for all writes using different length files, we can just check the file length. See PR #630 for the patch addressing these issues @bgaborg to get the changes I've done, look at the "command line instructions" at the bottom of the list, add my repo, do a git fetch and then cherry pick the commit 5aba1e4738b050fde1b2c45c21ed59320869be53 into this PR
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
