This commit adds the flexibility to configure kube apiserver qps
as per the user requirement and also keeps the existing values as
the default one.
Signed-off-by: yite.gu <yitegu0@gmail.com>
(cherry picked from commit 80541a23c9)
r is the variable name for the CSI reconciler
and same is used for a local variable as well
chaning it to avoid confusion and variable
shadowing.
Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
volumegroupsnapshot feature will be enabled
by default if the required CRD's are present
if not its disabled and user will have an option
to disable it if they dont require this feature.
Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
since networkFence is a cluster-based resource so that we don't need
the namespace and ownerReferences as it cause garbage-collector errors.
Also, now we create the networkFence with clusteUID label so when doing
cleanup we match the cephCluster uid and networkFence label clusterUID.
Signed-off-by: subhamkrai <srai@redhat.com>
This commit removes the `CSI_ENABLE_READ_AFFINITY` since
it is no longer utilized after addition of the read affinity
option per cluster via CSIDriverSpec.
Signed-off-by: Praveen M <m.praveen@ibm.com>
For now we are using the operator namespace name
as the prefix for the csi driver, This PR provides
an option for the users if someone wants to have
their own prefix for the csi driver, if someone tries
to change the prefix for existing csi driver rook
operator will fail to reconcile the csi driver.
Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
This commit adds the flexibility to configure
leader election flags as per the user
requirement and also keeps the existing values
as the default one.
resolve: #13475
Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
GRPC metrics got deprecated in cephcsi
3.7.0 and the deprecated flags will get
removed in the next release. This PR
removes the deprecated metrics code which
allow us to run with older cephcsi as well.
Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
when deploying csi, we'll check if csi-addons crds
are deployed we'll enable csi-addons sidecar by default.
Signed-off-by: subhamkrai <srai@redhat.com>
Adding maxUnavailable parameter to the cephfs csi daemonset. The
default value of maxUnavailable = 1 can make the upgrade process
slower in environments with large number of nodes.
Fixes: #12636
Signed-off-by: karthik-us <ksubrahm@redhat.com>
This commit removes k8s version check for oidc token
which required k8s v1.20 or above since minimum
k8s version that rook supports now is v1.22.
Signed-off-by: Rakshith R <rar@redhat.com>
adding a option to make AttachRequired field
as configurable and added a comment to explain
the consequences of changing the default value
resolves: #11805
Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
This commit adds code to supply variables required
to enable read affinity for rbd volumes.
It leverages `read_from_replica=localize,crush_locations=key:value`
krbd map option from ceph to serve reads from the most local OSD.
refer:
https://docs.ceph.com/en/latest/man/8/rbd/#kernel-rbd-krbd-options
Signed-off-by: Rakshith R <rar@redhat.com>
Rook will now append internal default versions to custom
csi image variables passed down if version is missing in
them.
This will enable users using their private registries to
just pass different image names while rook takes care
of adding default versions to them.
Resolves: #10529
Signed-off-by: Rakshith R <rar@redhat.com>
The default value of maxUnavailable of daemonset is 1.
Sometimes the upgrade can be slow due to this setting.
So I add a parameter to change the value for rbd daemonset.
Signed-off-by: Shinya Hayashi <shinya-hayashi@cybozu.co.jp>
Co-authored-by: Toshikuni Fukaya <toshikuni-fukaya@cybozu.co.jp>
Co-authored-by: Yuji Ito <llamerada.jp@gmail.com>
Introduce a new env variable ROOK_CSI_IMAGE_PULL_POLICY in rook operator configmap which should be used to
customize the imagePullPolicy for the csi driver and imagePullPolicy property in cephVersionSpec for ceph pods.
Signed-off-by: Avan Thakkar <athakkar@redhat.com>
On removing or setting params to false in operator config map, make sure
to set CSIParam to their appropriate defaults, since we are not restarting the
operator we cannot just assume these params/variables/flags are defaulted,
everytime we need to makesure some defaults are specified.
This is what happens without these changes:
* Set CSI_ENABLE_METADATA to true, this will tune CSIParam.CSIEnableMetadata
to true from false
* Create a PVC and check the metadata is added to the rbd/cephfs volumes
* Now Set CSI_ENABLE_METADATA to false, we assume CSIParam.CSIEnableMetadata
will be set to false, but that is not happening because the earlier value
of `CSIParam.CSIEnableMetadata = true` still persist.
(we are not restarting operator to assume value of CSIParam.CSIEnableMetadata
to be reset to false)
This happens with CSI_ENABLE_OMAP_GENERATOR and many other flags too,
fixing them all with these changes.
Thanks to Madhu for debugging the issue and testing it along.
Credit: Madhu Rajanna <madhupr007@gmail.com>
Signed-off-by: Prasanna Kumar Kalever <prasanna.kalever@redhat.com>
This commit adds topology provisioning
support. This makes modification to rbac,
csi deployment and daemonset.
Signed-off-by: Rakshith R <rar@redhat.com>
The volume replication operator is being moved to
kubernetes-csi-addons. This commit hence, removes
the volume replication sidecar and updates the
related documentation.
It will also update the csi-addons sidecar version
to the latest one.
Closes: #10655
Signed-off-by: yati1998 <ypadia@redhat.com>
Rook will support the most recent seven K8s releases,
so we update the min version to 1.19 for Rook v1.10.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
One can now use the operator configmap for CSI_CLUSTER_NAME variable
for setting ClusterName on the RBD images and CephFS subvolume.
This will be useful in cases like for example, when two container orchestrator
clusters (Kubernetes/OCP) are using a single ceph cluster.
Signed-off-by: Prasanna Kumar Kalever <prasanna.kalever@redhat.com>
One can now tweak the operator configmap for CSI_ENABLE_METADATA boolean to
enable/disable setting of PV/PVC/PVCNamespace &
snapshot-name/snapshot-namespace/snapshotcontent-name on the RBD images
and CephFS subvolume.
Signed-off-by: Prasanna Kumar Kalever <prasanna.kalever@redhat.com>
This commit introduces new log level called csidecarLogLevel
for the sidecars which are not maintained by the csi-driver.
This will allow to have different configuration options
for the sidecars like snapshotter, resizer, attacher and
provisioner containers which will help to avoid the
spamming of logs to an extent.
Signed-off-by: yati1998 <ypadia@redhat.com>
A Liveness sidecar container that is deployed
with the CSI pods will be helpful to check
csi driver is properly responding or not.
and there is no liveness or readiness
kubernetes probe added for this one.
Providing an option to enable this
sidecar deployment and disabling it by default.
fixes#8783
Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
moving setting csi parameter to setParams
function as `setParams` is mostly used to set
the csi parameters for the templates.
Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
This commit modifies ceph-csi controller to be able to cleanup
ceph-csi deployment,daemonset,services & csidriver objects
when no cephcluster exists.
It makes the following changes:
- redundant check for tp.DriverNamePrefix is removed.
- retry to start drivers within reconcile loop is removed,
controller will retry in case of error now.
- stopDrivers() will now return error in case of failure.
- predicate delete func will now respond to cephcluster deletion.
- CSI resources will be cleaned up when cephcluster does not exist.
Fixes: #9697
Signed-off-by: Rakshith R <rar@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>
This commit moves SetParams and ValidateCSIParam func
into go routine/lock, since it modifies/reads global CSIParam
variable.
Signed-off-by: Rakshith R <rar@redhat.com>
This commit changes CSI_ENABLE_HOST_NETWORK default value to true
since it was observed that cephfsvolume gets blocked when csi cephfs
nodeplugin is restarted and nodeunpublish call also hangs when
cephfs nodeplugin is using pod networking.
Updates: #8085
Signed-off-by: Rakshith R <rar@redhat.com>
CSI drivers failed to start if rook encountered error
while reading configmap with no retry.
This commit introduces retry for starting CSI driver
if there is a failure in reading configmap upto 3 times.
Fixes: #7950
Signed-off-by: Rakshith R <rar@redhat.com>
As host networking is no longer necessary for CSI driver,
it will be disabled by default.
An option CSI_ENABLE_HOST_NETWORK is added to rook-ceph-operator-config
configmap and enableCSIHostNetwork in helm chart values.yaml to
enable/disable host network in CSI plugin pods.
Closes: #7203
Signed-off-by: Rakshith R <rar@redhat.com>
It's better to validate ownerReferences when setting them. In addition, we should use
controllerrutil.Set{Controller,Owner}Reference, that have such validation, as possible.
Signed-off-by: Satoru Takeuchi <satoru.takeuchi@gmail.com>
The GRPC metrics exposed by both the provisioner pod
and the node plugin pod on some port. Provisioner pod
is running on the pod network and the daemonset pods
run on the host network. sometimes starting the GRPC
metrics by default can lead to node plugin pods
crashloopback state this is due to the port conflict.
Moreover, the GRPC metrics are not for the user it's
for the one which will help to debug the time taken
by cephcsi to serve each GRPC call. Enabling it by
default won't be a good idea. So the plan is to
disable it by default, If someone faces any issue it
can be enabled later at some point in time.
closes#7378
Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
Before this patch, csi detection will run whether or
not rbd and cephfs are all disabled. Run detection only
if it is necessary.
Signed-off-by: shenjiatong <yshxxsjt715@gmail.com>
Support ceph cluster and CSI on multus deployed in diffrent namespace.
Previously csi was looking for multus config from the cluster deployed
in the namespace in which rook-ceph-operator/csi was deployed.
Now it will look for multus configuration from ceph clusters from all
the namespaces.
Signed-off-by: rohan47 <rohgupta@redhat.com>
This adds the functionality to add custom pod labels to the CSI
components through the operator configuration way of env vars or config
map.
Resolves#6593
Signed-off-by: Alexander Trost <galexrt@googlemail.com>
If the AllowUnsupported flag is set to true we should
not check the csi version as it will be kind of no-op
as even incase of failure we continue. if we disable
the version check in this case,the deployment will
be little faster as we dont wait for version check
to complete to start CSI pods.
Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
remove csi drivers and delete k8s
services that drivers create,when
the default setting of drivers changed
to false.
Signed-off-by: subhamkrai <subhamkumarrai03@gmail.com>
The csi driver does not need to be started before the cluster
is created. When the csi driver fails to load, neither should
it prevent the cluster from being created. Therefore, we
initialize the basic constructs in the csi driver that are required
for cluster creation, then start the csi driver in a goroutine
so the cluster creation can continue. If the csi driver fails
or takes a long time to load, it will no longer affect cluster
creation.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
yite.gu
Madhu Rajanna
subhamkrai
Praveen M
karthik-us
Rakshith R
Avan Thakkar
Prasanna Kumar Kalever
yati1998
Travis Nielsen
Sébastien Han
bipuladh
Satoru Takeuchi
shenjiatong
Arun Kumar Mohan
rohan47
Alexander Trost
subhamkrai