This is not an issue affected by turning lookup-optimize=on. This is for the reason that its an add-brick scenario, during which clients resort to lookup-everywhere even though lookup-optimize=on. Hence dropping the tag.
This issue is not seen with glusterfs version 3.7.9-10.el7rhgs.x86_64.
Here are the steps that were performed,
1. Created a distributed replica volume and mounted it.
2. Created Directories from mount point.
4. Added new bricks to this volume. Note: Did not start rebalance here.
5. From mount point executed commands which look for those directories.
6. Verified hash layout for those directories on newly added bricks.
Directory hash layout on the newly added brick are set to zeros and the layout for those directories are only fixed when performed rebalance.
Comment 4Nithya Balachandran
2016-09-01 08:58:43 UTC
Thanks Prasad. Moving this to ON_QA based on comment#3.