Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
CA-390490: remove override of blktap request timeout
When this code was orginally added there was no default value for the nfs_timeout and so the timeout would only have been used if 1. The SR type was NFS 2. The SR had a overriden setting for nfs-timeout in the other-config Since that time the calculation of the NFS client timeout has been changed considerably and is set with a balance between timeo and retrans values. In particular the get_nfs_timeout method will now always return a non-zero value. This results in the SM always applying a request timeout to blktap of 50 seconds (unless the timeo value has been set in the SR device config), which is considerably lower than the default request timeout in blktap itself which is 120 seconds. NFS protocol timeouts are also no longer considered to be retryable errors within blktap so tying the timeout value to the NFS timeout, even if it were correctly calculated taking into account the retrans value, is effectively comparing two unrelated things. Signed-off-by: Mark Syms <[email protected]>
- Loading branch information