Description of problem: Seeing "transport endpoint error" on copying files when the the hot tier in unavailable Version-Release number of selected component (if applicable): glusterfs-server-3.7dev-0.885.git0d36d4f.el6.x86_64 Volume Name: tiervol Type: Tier Volume ID: 3744cb8f-c6a2-4d31-b034-9c590989118a Status: Started Number of Bricks: 2 x 1 = 2 Transport-type: tcp Bricks: Brick1: 10.70.35.166:/b5 Steps to Reproduce: 1. Create a simple distributed volume 2. Attach another distribute (as hot tier) to the created volume 3. Mount the volume on the client 4. Kill the brick process for the hot tier 5. Initiate copy of data in the mount 6. Seeing "Transport endpoint is not connected" error on copy Expected Result: Even if the hot tier is down, the copy should continue on the cold tier.
This bug is unclear. We need to know if the file resides on the hot tier or the cold tier. If the file existed on the hot tier, it will not be available when you kill the brick process. A file only exists on one tier, not both. If the file exists on the cold tier, it would be available, and we would chase it. Probably the best way to document this would be including in the bug ls /brick_hot and ls /brick_cold We can discuss more to drill down. For now I'll put this bug in ON_QA pending our discussion.
This is for fresh writes. All fresh writes are failing when hot tier is unavailable. Reproduced this ont the BETA2 build too, hence moving it to ASSIGNED.
Even if it is a fresh write we cannot create a new file. We cannot know if the file already exists on the hot tier. If it turned out the file did already exist, and we created a second file with the same name on the cold, this would overwrite the old file and be dataloss. This is not a bug.