Bug #23326
closed
mgr: Error EIO: Module 'dashboard_v2' has experienced an error and cannot handle commands: No module named rbd
Added by Patrick Donnelly about 6 years ago.
Updated about 3 years ago.
Description
When running vstart against master:
pdonnell@senta02 ~/ceph/build$ ../src/vstart.sh -d -b -l -n
...
Starting mgr.x
/home/pdonnell/ceph/build/bin/ceph-mgr -i x -c /home/pdonnell/ceph/build/ceph.conf
/home/pdonnell/ceph/build/bin/ceph -c /home/pdonnell/ceph/build/ceph.conf -k /home/pdonnell/ceph/build/keyring tell mgr restful create-self-signed-cert
no valid command found; 10 closest matches:
mgr self-test run
prometheus self-test
influx config-show
influx config-set <key> <value>
influx self-test
influx send
balancer execute <plan>
balancer dump <plan>
dashboard set-session-expire <int>
dashboard set-login-credentials <username> <password>
Error EINVAL: invalid command
MGR Restful is not working, perhaps the package is not installed?
/home/pdonnell/ceph/build/bin/ceph -c /home/pdonnell/ceph/build/ceph.conf -k /home/pdonnell/ceph/build/keyring mgr module enable dashboard_v2
/home/pdonnell/ceph/build/bin/ceph -c /home/pdonnell/ceph/build/ceph.conf -k /home/pdonnell/ceph/build/keyring tell mgr dashboard set-login-credentials admin admin
Error EIO: Module 'dashboard_v2' has experienced an error and cannot handle commands: No module named rbd
Generally that's what you'd get from an incomplete build, perhaps doing the `cephfs_testing` vstart target rather than the `vstart` one?
Oh, I'm building with:
./do_cmake.sh -DWITH_BABELTRACE=OFF -DWITH_MANPAGE=OFF -DWITH_RBD=OFF -DWITH_RADOSGW=OFF -DWITH_TESTS=OFF && time (cd build && make -j24 CMAKE_BUILD_TYPE=Debug -k)
Omitting RBD must be why. Is there a way I can avoid this mgr error without building RBD?
Patrick Donnelly wrote:
Omitting RBD must be why. Is there a way I can avoid this mgr error without building RBD?
Currently no. dashboard_v2 has a hard dependency on RBD, therefore you need to built it.
It's not impossible to put some special handling in the dashboard for disabling rbd functionality when it's missing, but I'm not sure it's worth the effort compared with just compiling with rbd?
Patrick: is it okay with you to "Won't fix" this?
I understand the temptation but I'd prefer this be fixed. Taking out RBD compilation has significantly improved compilation time for me when testing changes (especially on my laptop with limited cores).
I think what's going on here is that the dashboard used to fail silently (it always required rbd bindings), because vstart was enabling it via "mgr initial modules" and then never trying to use any commands on it. So what you probably need is a vstart flag to just disable the dashboard, to reinstate the effective previous behaviour - right?
Still, I'd suggest usually taking the hit of building RBD (at least if you're running somewhere other than a laptop) so that you can use the cephfs dashboard bits on vstart clusters.
John Spray wrote:
I think what's going on here is that the dashboard used to fail silently (it always required rbd bindings), because vstart was enabling it via "mgr initial modules" and then never trying to use any commands on it. So what you probably need is a vstart flag to just disable the dashboard, to reinstate the effective previous behaviour - right?
A vstart flag works for me.
A vstart flag is easier to implement, and we should fix this problem that way in the short term. But we can also think of dynamically disabling some of the dashboard features if the required dependencies are not available in the long term.
- Status changed from New to Resolved
- Assignee set to Joao Eduardo Luis
- Project changed from mgr to Dashboard
- Category changed from 132 to General
Also available in: Atom
PDF