The uid/gid is determined by looking at file ownership. It
was looking at /etc/ceph, but on current builds this is owned by root so switched to using /var/lib/ceph which
is the homedir of the ceph account.
Signed-off-by: Paul Cuzner <pcuzner@redhat.com>
Initial support to enable cephadm to deploy
monitoring containers like prometheus. This
patch adds support for prometheus. It is the callers
responsibility to provide a valid prometheus.yml
file passed within the --config-json parameter.
Signed-off-by: Paul Cuzner <pcuzner@redhat.com>
* refs/pull/32158/head:
cephadm: fix exception when no time sync is running
cephadm: properly extract osd fsid during adoption
cephadm: do ceph-volume activate+deactivate as part of systemd unit
ceph-daemon: behave with no /etc/ceph/ceph.conf
Reviewed-by: Jan Fajerski <jfajerski@suse.com>
Reviewed-by: Sebastian Wagner <swagner@suse.com>
This needs to be passed to the deploy_daemon_units method so that we can
set up the pre and post c-v calls properly.
Signed-off-by: Sage Weil <sage@redhat.com>
1- Do the activate as part of the unit on start, since it may include
setup that disappears on a reboot, like starting up the dmcrypted volume.
2- Do the matching deactivate step when the container stops, like shutting
down the dmcrypt volume.
Right now the OSD is the only thing that needs this special behavior.
Signed-off-by: Sage Weil <sage@redhat.com>
* refs/pull/32192/head:
common: handle return value from read(2)
Reviewed-by: Jeff Layton <jlayton@redhat.com>
Reviewed-by: Sage Weil <sage@redhat.com>
* refs/pull/32195/head:
osd/PeeringState.cc: skip peer_purged when discovering all missing
Reviewed-by: Sage Weil <sage@redhat.com>
Reviewed-by: David Zafman <dzafman@redhat.com>
* refs/pull/32144/head:
doc: mention multimds suite in dev guide
doc: clarify difference between fs and kcephfs suite
Reviewed-by: Varsha Rao <varao@redhat.com>
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
/usr/bin/python dne on el8, /usr/bin/python3 dne on el7. But
all we need to do is write a file--we can do that with tee.
Signed-off-by: Sage Weil <sage@redhat.com>
The problem was that the TSLint linting hasn't matched the
linting of prettier regarding the quotes of a string.
Explanation:
'\'' will be converted to "'" with prettier, but TSLint thrown an
error that it wanted the string to look like '\''.
Now TSLint will not error if it sees an single quote inside a double
quote to match the behavior of prettier.
Fixes: https://tracker.ceph.com/issues/38287
Signed-off-by: Stephan Müller <smueller@suse.com>
Adds simple validators that resemble the min and max validators for
numbers.
As a binary size (the string) has first be converted into bytes
(the number) to validate against some other number and in case of an
error the other number want's to be converted into a size to display,
the whole process needs a bunch of conversion, but these validators take
care of all that.
Fixes: https://tracker.ceph.com/issues/38287
Signed-off-by: Stephan Müller <smueller@suse.com>
Now any input type is supported and the special binary type which will
use the cd-binary input and automatically format the binary size into
bytes when submitting.
Now any field can have custom validators and error messages.
The form will fallback on predefined error messages if not configured.
It will use the error messages provided by the binary min and max
validators. It still provides the error message for a required field.
Fixes: https://tracker.ceph.com/issues/38287
Signed-off-by: Stephan Müller <smueller@suse.com>
Now both CephFS quotas can be changed with a validation against the next
tree maximum in the file tree, that prevents setting the quotas in a way
that would not be usable.
Fixes: https://tracker.ceph.com/issues/38287
Signed-off-by: Stephan Müller <smueller@suse.com>
When stop.sh is run while the cephFS is mounted,
mount processes are retained and can't be killed.
And also the mount and df commands hang after this.
This patch unmounts the mounted volumes cleanly.
Signed-off-by: Kotresh HR <khiremat@redhat.com>
