In K8s 1.22 there is a bug in the job name generation that
the job name is truncated an additional 10 characters. This can cause an issue
in the generated pod name if it then ends in a non-alphanumeric character. In that case,
we more aggressively generate a hashed job name.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
The failover of the arbiter mon in a stretch cluster was sometimes
failing due to the new tiebreaker not being set in ceph.
Rook would repeatedly try to remove the old tiebreaker mon
and keep failing because the new tiebreaker had not been set.
Now we make setting the tiebreaker idempotent in case the operator
restarts in the middle of the operation or some other corner
case causes the expected tiebreaker to be set. In that case,
the next reconcile will also ensure the tiebreaker mon is
set as expected.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
Skipping upgrade checks was not being honored for OSDs.
Now the flag will be checked and allow the OSDs to be upgraded
without checking for the ok-to-stop condition.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
If we apply useAllNodes to false for the current deployment,
the OSDs should get updated with the individual nodes values and config,
The deviceClass was not updating to the existing OSDs because there was
bug in the check.
The check osdInfo.DeviceClass == "" which should be
checked like this osdInfo.DeviceClass == "None"
Updated the code so OSDs can make use of the devices present
Signed-off-by: parth-gr <paarora@redhat.com>
The cluster info is important context for the cluster controller to
create the cluster, and all the fields must be properly set.
A test cluster name was being set temporarily, resulting in
mons incorrectly getting the wrong cluster CR name. There is no
known issue from the temporary value, it was just exposed by
https://github.com/rook/rook/pull/8678 setting the value to a label.
Now the functions are more clearly named so only unit and
integration tests should be using the test value for the cluster
name where it is not important.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
the mon cluster clusterInfo is intiated seprately,
and misses out to set the cluster name and use default name as testing
from AdminClusterInfo.
Part-of: https://github.com/rook/rook/issues/9159
Signed-off-by: parth-gr <paarora@redhat.com>
The templates for the mgr-generated ServiceMonitor and PrometheusRule objects included the labels
prometheus and team, making it impossible to override them as user.
This adds a new method `OverwriteApplyToObjectMeta` to
pkg/apis/ceph.rook.io/v1.Labels, which, contrary to the existing
`ApplyToObjectMeta` method, overwrites existing labels.
Closes: https://github.com/rook/rook/issues/8502
Signed-off-by: Mara Sophie Grosch <littlefox@lf-net.org>
The blockdevmapper securityContext was changed to request a minimal set of
required capabilities for its operation and drop running as privileged.
While the base change works and is valid in terms of the container's copy operation,
it turns out that OpenShift may require some additional configuration not
currently covered by the limited securityContext and the capabilities granted.
To not break those OpenShift deployments, make the blkdevmapper securityContext
listen to the ROOK_HOSTPATH_REQUIRES_PRIVILEGED flag again to set privileged mode.
This flag is true on OpenShift deployments and running as privileged
works around the (missing) configuration problem for now.
To properly drop privileged completely some additional investigation needs
to be done on OpenShift deployments without relying on privileged execution.
Signed-off-by: Omar Pakker <Omar007@users.noreply.github.com>
From ceph v16.2.6 onwards the vault TLS suppport in RGW was added,
include similar changes for RGW.
Signed-off-by: Jiffin Tony Thottan <thottanjiffin@gmail.com>
This commit adds context parameter to k8sutil node functions. By this,
we can handle cancellation during API call of node resource.
Signed-off-by: Yuichiro Ueno <y1r.ueno@gmail.com>
This commit adds context parameter to k8sutil job functions. By this, we
can handle cancellation during API call of job resource.
Signed-off-by: Yuichiro Ueno <y1r.ueno@gmail.com>
the mons re-initialize its ClusterInfo which results in missing
CR name on the ClusterInfo
Set the CR name to the mons cluster ClusterInfo
Closes: https://github.com/rook/rook/issues/9159
Signed-off-by: parth-gr <paarora@redhat.com>
The OSD blkdevmapper init container relies on the MKNOD capability,
which it does not actually request.
As a result, deployments fail on Kubernetes clusters that do not
happen to assign this capability to all containers by default.
Solve this by updating the container spec securityContext to
explicitly request the capability it relies on.
Closes: https://github.com/rook/rook/issues/9156
Signed-off-by: Omar Pakker <Omar007@users.noreply.github.com>
This commit adds context parameter to k8sutil pod functions. By this, we
can handle cancellation during API call of pod resource.
Signed-off-by: Yuichiro Ueno <y1r.ueno@gmail.com>
This commit adds context parameter to k8sutil deployment functions. By
this, we can handle cancellation during API call of deployment resource.
Signed-off-by: Yuichiro Ueno <y1r.ueno@gmail.com>
When a reconcile is started for OSDs, the prepare jobs are first
deleted from a previous reconcile. The timeout for the osd prepare
job deletion was only 40s. After that timeout, the reconcile attempts
to continue waiting for the pod, but of course will never complete
since the OSD prepare was not running in the first place, causing the
reconcile to wait indefinitely. In the reported issue, the osd prepare
jobs were actually deleted successfully, the timeout just wasn't long
enough. Pods need at least a minute to be forcefully deleted,
so we increase the timeout to 90s to give it some extra buffer.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
In the event a ceph image is specified that is lower than the current running
version of the daemons, the downgrade is allowed, even if not technically
supported. All of the core daemons (mon,mgr,osd) were being downgraded,
but the daemons for other controllers (rgw,mds,rbdmirror) were not being
downgraded, resulting in an inconsistent cluster. Now we log that the downgrade
is not supported and all all of the daemons to be downgraded.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
On Pacific, Ceph's default is "upmap", so we should let it be
like this.
This lets the user change the mode is desired.
On Octopus though, Rook continues to force the mode to "upmap".
Closes: https://github.com/rook/rook/issues/9062
Signed-off-by: Sébastien Han <seb@redhat.com>
The cluster CR can be forcefully deleted and cleanup the
cluster resources if the yes-really-destroy-data policy
is set on the CR. In this case, the other controllers should
treat the cluster CR as not existing and allow the finalizers
to be removed on those resources if they are requested for
deletion.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
We can now set labels to the crash collector deployment by editing the
CephCluster CR with:
```yaml
spec:
labels:
crashcollector:
```
Closes: https://github.com/rook/rook/issues/9039
Signed-off-by: Sébastien Han <seb@redhat.com>
This commit adds context parameter to utilities in opcontroller to
remove context.TODO use in opcontroller. By this, we can handle
cancellation of reconcilers in a fine-grained way.
Signed-off-by: Yuichiro Ueno <y1r.ueno@gmail.com>
Rook cluster-wide encryption can now use the native Kubernetes
authentication to interact with vault KMS instead of using the token
method.
Signed-off-by: Sébastien Han <seb@redhat.com>
When deploying a cluster-wde encrypted cluster we now use the rook
binary to execute some code to fetch the key encryption key.
Using cURL all the time to fetch the key has its limitations. The
incoming integration with Kubernetes Authentication through service
accounts is leading the usage of cURL to its end.
The logic is really complex and prone to errors. Re-implementing the
Go logic into Bash is not a viable option.
Also, using the lib in a binary allows us to keep a consistent behavior
throughout the life cycle of our code.
Signed-off-by: Sébastien Han <seb@redhat.com>
Prior to ceph v16.2.7 the failover of the arbiter mon was
not supported. Now the new tiebreaker mon can be set during
the failover event and provide more dynamic stability to
the mon quorum if another node is available in the arbiter
zone.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
Adding finalizers to rook-ceph-mon secrets
and rook-ceph-mon-endpoints configmap
We don't want to delete this resources during disaster
because these details are needed during disaster recovery
Closes: https://github.com/rook/rook/issues/8369
Signed-off-by: parth-gr <paarora@redhat.com>
We don't need to use tini.
We don't have anything in the rook operator that would
either create zombie processes (no threads) or use
exec (to fork). The Go binary has a really good
signal handling mechanism.
Closes: https://github.com/rook/rook/issues/8794
Signed-off-by: Sébastien Han <seb@redhat.com>
The flex driver has been fully deprecated and thus removed from Rook.
Before upgrading to v1.8, users will need to convert existing flex volumes
from flex to csi volumes.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
When private/public network have been defined in the Ceph rook cluster it is not
possible to configure properly the ip address of the liveness probe
for the manager.
Changes in the manager in Pacific introduced this regression.
This change replaces the http probe by a command probe, avoiding thus to
determine what is going to be the ip address of the manager before launching
the pod.
fixes: https://github.com/rook/rook/issues/8510
Signed-off-by: Juan Miguel Olmo Martínez <jolmomar@redhat.com>
This commit is a large refactor on how the operator starts, stops and
how it starts various sub-components such as the ceph-csi driver. It
also refines the way we cancel orchestrations. We don't use breakpoints
anymore but send our self a SIGUP to reload our controller runtime
manager.
The reload will happen under different circonstances like:
* a new adminission controller secret is created/deleted/changed
* a CephCluster CR is edited
As mentioned earlier, the csi driver now has its own controller, just
like flex. It reacts to change in the operator config map for particular
ROOK_CSI_ fields.
A second new controller for the operator's general config has been
created, it manages:
* the logging level
* the ceph CLI command timeout
* the discovery daemon
The operator reacts much more rapidly to cancellation events by stopping
the manager's context and reloading it.
Signed-off-by: Sébastien Han <seb@redhat.com>
correct typo in logging, it was showing `ok-to-stop`
instead of `ok-to-continue` when 'continueUpgradeAfterChecksEvenIfNotHealthy' is true
Co-Authored-by: Zeaone <zeaone@ZeaonedeMacBook-Pro.local>
Signed-off-by: subhamkrai <srai@redhat.com>
While an even number of mons can cause lower availability of
mon quorum, it also can provide higher durability for the cluster.
Mon quorum can be restored from a single mon according to the
disaster recovery guide, so there may be scenarios where
an even number of mons may be preferable.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
Travis Nielsen
parth-gr
Blaine Gardner
Mara Sophie Grosch
Omar Pakker
Jiffin Tony Thottan
Sébastien Han
Yuichiro Ueno
Yuval Lifshitz
subhamkrai
Juan Miguel Olmo Martínez
JrCs