Description of problem: setup: replica x3 no distribution using encrypt xlator. File read returns some blocks corrupted. The chance of appearance for this corruption increases as file size gets bigger. So, there is higher probability for the issue to be visible in bigger files. Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. Copy a file in an encrypted volume (most preferably a file of >20MB size for higher probability) 2. From a client, take the MD5 checksum of that file 3. Every execution of MD5 calculation returns a different hash Actual results: File returned in read operations is corrupted. Expected results: File should not be corrupted. Additional info: The issue seems to be a bug during concurrent access in encryption translator during XTS encryption. At all times the problem appears to one or two blocks of the file while the others are fine. The reason seems to be related to the initialization vector that xlator uses for the XTS encryption.
This bug is moved to https://github.com/gluster/glusterfs/issues/851, and will be tracked there from now on. Visit GitHub issues URL for further details