Description of problem: Problem: ------- While the glfs object of a volume is cached already, if you delete the volume and create the volume with the same name, gluster-block currently uses the old volumes glfs object, which is not expected. Why? Because, current lru cache Hashing is based on volume name. Solution: -------- we would need to discard a cached object upon some number of failures or an option to explicitly discard an object from cache, or some equivalent policy
*** Bug 1594112 has been marked as a duplicate of this bug. ***
> Why? > Because, current lru cache Hashing is based on volume name. Considering this is not expected to happen in a OCS scenario, would be great to take this to Upstream, and fix it there instead if valid. Also noticed that the other bug Niels said he is not able to reproduce the problem.