ceph_context: re-expand admin_socket metavariables in child process #18393
Conversation
|
retest this please. |
|
it's a general problem for all ceph daemons. is it possible to fix problem completely? |
Reset admin_socket raw value if it is defined in conffile. Just in case it used metavarirables (e.g, $pid) which could be expanded again to the correct value in child process. Signed-off-by: Zhi Zhang <zhangz.david@outlook.com>
david-z
force-pushed
the
wip-ceph-fuse-admin-socket
branch
from
9fcff4a
to
1630f4b
Compare
david-z
changed the title
|
@ukernel I checked again on ceph-mon, ceph-osd, ceph-mds, ceph-mgr and ceph-fuse daemons. Except ceph-fuse, other daemons don't have this problem right now. But if a daemon is started in a child process and the value of its "daemonize" option is true, this problem would happen. So once for all, I moved the fix into ceph_context just before creating admin_socket. Please review the new changes, thanks. |
* refs/pull/18393/head: ceph_context: re-expand admin_socket metavariables in child process Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
|
This change introduces a regression: admin-socket, specified via command line or CEPH_ARGS env does not work: E.g. this test fails now: qa/workunits/rbd/test_admin_socket.sh |
|
I mean, if you have admin-socket specified both in ceph.conf and via command line (or env), it will use ceph.conf setting instead of command line override. |
|
Another thing that I don't like very much about this solution is that it tries to fix only "admin-socket" config option, while $pid can be useful in other options too (e.g. log-file). |
Reset admin_socket raw value if it is defined in conf file. Just in case it used metavarirables (e.g, $pid) which could be expanded again to the correct value in child process.
Fixes: #http://tracker.ceph.com/issues/21848
Signed-off-by: Zhi Zhang zhangz.david@outlook.com