[DRBD-user] drbdadm status could tell more about errors
Håkan T Johansson
f96hajo at chalmers.se
Mon Nov 21 08:55:36 CET 2022
I am using drbd-utils from Debian bullseye (9.15.0-1) and the 9.2.1 kernel
module (local compile). I'm new to drbd and made a mistake, and would
just like to suggest that 'drbdadm status' tries to help more with errors:
- First I had tested with the 8.4.11 kernel module between two hosts.
- Then I wanted to try the new quorum system, so got the newer kernel
module and added a third node. However, without being careful, the
partition I choose on that machine was smaller than the original two.
- At first connections worked, but after having forcefully done some
'drbdadm create-md' then the system apparently realised the size
mismatch and refused further connections.
Here 'drbdadm status' basically reported 'Connecting' for the connections
between the third host and the other two.
This sounded like networking issues, but with tcpdump I could see messages
actually exchanged, and only after finding some log messages of this kind:
"Considerable difference in lower level device sizes: 586078856s vs.
31249136s" I figured my partition sizing mistake.
It would be nice if 'drbdadm status' would be more explicit in the error
status in this case, e.g. say something like 'Connecting (Refused;
Size mismatch)'. Where 'refused' would indicate that network exchange has
been successful, and 'size mismatch' why connection was not established.
In general, the 'xxx connection:Connecting' line could tell more about how
far the last connection attempt got.
Sorry for the noise if this has already been adressed in a newer version.
More information about the drbd-user