REVIEW: http://review.gluster.org/9343 (gfapi: new inode created in glfs_resolve_component even if inode is in inode table) posted (#1) for review on release-3.6 by Sachin Pandit (spandit)
REVIEW: http://review.gluster.org/9343 (gfapi: new inode created in glfs_resolve_component even if inode is in inode table.) posted (#3) for review on release-3.6 by Sachin Pandit (spandit)
commit 6b83a658ede66ebbb0983f1e4a326fd12cc91c34 Author: Rajesh Joseph <rjoseph> Date: Mon Dec 8 14:25:22 2014 +0530 gfapi: new inode created in glfs_resolve_component even if inode is in inode table problem: USS allows split-brain file to be accessed while main volume gives I/O error. cause: AFR detects split-brain on lookup. It stores this information in inode context. open and readv fop checks this flag from inode context. open and readv fop fails if split-brain flag is set for the file. USS uses gfapi to access snapshot volume. During open call gfapi internally calls glfs_resolve_component. glfs_resolve_component generates a new inode even if inode is present for the file. Because of which afr_lookup acts on a new inode which does not contain the split-brain flag. Bug: 1171615 Change-Id: I1b4fddf4bd3c734a319ecfae804a3439866d157c Signed-off-by: Rajesh Joseph <rjoseph> Reviewed-on: http://review.gluster.org/9253 Reviewed-by: Vijaikumar Mallikarjuna <vmallika> Reviewed-by: Raghavendra Talur <rtalur> Reviewed-by: Poornima G <pgurusid> Reviewed-by: Pranith Kumar Karampuri <pkarampu> Tested-by: Gluster Build System <jenkins.com> Reviewed-by: Niels de Vos <ndevos>