You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
These days there are udev rules and possibly other mechanisms by which arrays may already be assembled by the time someone or some script (perhaps an init script) runs mdadm --assemble --scan. Problems arise when mdadm subsequently exits with status 2, because the caller will handle that as a failure even though from its perspective everything is actually fine. In an init script, for example, this may cause dependent init scripts to fail to start.
Furthermore, while exit statuses are documented for some of the other modes, they do not appear to be documented for --assemble mode. If for some reason the exit status for this case cannot be changed to 0, then please at the very least exit with some unique documented status that callers can check for specifically.
Here are a couple of examples of downstream bug reports due to this behavior:
Assemble implementation is hard to follow. I think that it would be easier and safer to write new assemble module instead.
Can you propose documentation that would satisfy your needs?
These days there are udev rules and possibly other mechanisms by which arrays may already be assembled by the time someone or some script (perhaps an init script) runs
mdadm --assemble --scan
. Problems arise whenmdadm
subsequently exits with status 2, because the caller will handle that as a failure even though from its perspective everything is actually fine. In an init script, for example, this may cause dependent init scripts to fail to start.Furthermore, while exit statuses are documented for some of the other modes, they do not appear to be documented for
--assemble
mode. If for some reason the exit status for this case cannot be changed to 0, then please at the very least exit with some unique documented status that callers can check for specifically.Here are a couple of examples of downstream bug reports due to this behavior:
The text was updated successfully, but these errors were encountered: