On Thu, Dec 14, 2023 at 06:26:58PM +0100, Martin Wilck wrote: > On Tue, 2023-12-12 at 18:53 -0500, Benjamin Marzinski wrote: > > Now that multipathd is running the same code to remove devices as > > multipath, multipath doesn't need to automatically retry the remove > > if > > it fails. > > > > Signed-off-by: Benjamin Marzinski <[email protected]> > > If we know that multipathd executes the same commands that multipathd > does, why don't we just call delegate_to_multipathd() repeatedly until > the retries are exhausted?
We could. The benefit of the multipath command is a more specific error when the device is in use, and it will work even if multipathd is wedged. On the other hand users can always check the logs for the specifics of which device was in use and we could handle ETIMEDOUT or a reply of "fail\ntimeout" specially. We might want to do that even if the user doesn't ask for retries, since it would be useful for all the delegated commands to failback to doing the work in multipath if multipathd is timing out. -Ben > Martin
