When writing to the operator log, where possible
the namespace of the resource on which the operator
is working will be logged as a standard prefix
to more easily troubleshoot what resource is being reconciled
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
Labels cannot be longer than 64 characters, and if the image name
is too long we log a verbose message about the label that we can
just log now as debug level.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
With a osd resize operation, currently pvc and pv may
takes more time then expected.
And so the osd dont get resized and need manual re-start
as currently we don't have a watch on osd's
controller that watches pv and get it resized.
add a wait time till pvc is not resized
Signed-off-by: parth-gr <partharora1010@gmail.com>
there are some cases where storageClassDeviceSets length
are more 63 chars and osd provisioned failed due to this.
with error
```
failed to run osd provisioning job for PVC "ocs-deviceset-*-local-volume-storageclass-0-data-*": Job.batch "rook-ceph-osd-prepare-*" is invalid: [spec.template.spec.volumes[10].name: Invalid value: "ocs-deviceset-*-local-volume-storageclass-0-data-*": must be no more than 63 characters, spec.template.spec.containers[0].volumeMounts[9].name: Not found: "ocs-deviceset-*-local-volume-storageclass-0-data-*",
```
Using x-validator to limit the maxLength to the deviceClassSet.Name to
40 chars and adding code check that overall pvc name should be less than
63 chars.
Signed-off-by: subhamkrai <srai@redhat.com>
Reduce CRD size by removing some descriptions for things that
(a) are long, (b) are repeated, and (c) are commong kubernetes types.
Remove the `Status` field from VolumeClaimTemplates (a.k.a. PVCs) in
types.go to remove all unnecessary status-related fields.
Use `yq` to go through generated `crds.yaml` to strip out descriptions
for well-known k8s types that are long and repeated where we can't
remove them from types.go. This is mostly placement descriptions and
volumeSource descriptions.
The overall `crds.yaml` file is reduced to less than 60% of its original
size.
Signed-off-by: Blaine Gardner <blaine.gardner@ibm.com>
Signed-off-by: Redouane Kachach <rkachach@redhat.com>
Co-authored-by: Blaine Gardner <blaine.gardner@ibm.com>
This PR makes Rook operator to add a label on PVC that contains an image version of Ceph when creating an OSD.
Signed-off-by: YZ775 <yuzuki-mimura@cybozu.co.jp>
The volume names of a pod must follow the standard naming convention
which does not allow periods in the name. Since the device set name
is used as the pvc name and then the volume name to attach to the osd
prepare job, we must sanitize the name for the invalid character
and replace it.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
The purge job is intended to clean up all the resources
related to an OSD that is to be removed. An OSD with
a metadata, wal, and data PVC was only cleaning up the
data PVC. Now the metadata and wal PVCs will also be
cleaned up.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
This commit adds context parameter to k8sutil pvc functions. By this, we
can handle cancellation during API call of pvc resource.
Signed-off-by: Yuichiro Ueno <y1r.ueno@gmail.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>
The rook.io/v1 package was only an internal implementation detail and
does not have any CRDs that rely on it. The CRD deserialization should
handle the change in internal types without any issue. This separation
gives more flexibility for the storage providers to implement exactly
what is needed for their storage provider instead of forcing to use the
same types and risk affecting another storage provider.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
A user with unbalanced system (e.g., some OSDs use a partition which
resides on a device along-side OS partition) may want to have a
fine-grained control on the overall load of specific OSD. While osd
weight affects write-load, primary-affinity affects read-load.
This patch allow user to define osd's primary-affinity (similar to
'DeviceClass' and 'InitialWeight'), which is sent to the osd (via mon)
as part of deployment sequence, using 'ceph osd primary-affinity' command.
Valid values are within the range [0.0, 1.0). If not set, default value
is 1.
ROOK issue: https://github.com/rook/rook/issues/7773
Signed-off-by: Shachar Sharon <ssharon@redhat.com>
The VolumeSource type was only used internally when loading
the storageClassDeviceSets. The volumeSource element is never
expected to be specified in the CephCluster CR, so it is now
removed from the schema. The VolumeSource type is renamed to
deviceSet for clarity.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
delete updatePVCIfChanged method on "Cluster" from osd package and use
the common one ExpandPVCIfRequired from k8sutil pkg
Signed-off-by: Nitin Goyal <nigoyal@redhat.com>
Ceph support the option '--osd-crush-initial-weight' upon OSD start,
which sets an explicit weight (in TiB units) to specific OSD. Allow
passing this option all the way from the user (similar to
'DeviceClass'), for the special case where end users wants it cluster
to have non-even balance over specific OSDs (e.g., one of the OSDs is
placed over a partition alongside OS-partition).
ROOK issue: https://github.com/rook/rook/issues/7448
Signed-off-by: Shachar Sharon <ssharon@redhat.com>
The metadata name must be unique for a storageClassDeviceSet since
only one each of the sections data, metadata, and wal can be specified.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
Update OSDs in parallel per the design in
design/ceph/update-osds-in-parallel.md
The max number of OSDs updated in parallel is currently fixed at 20.
Signed-off-by: Blaine Gardner <blaine.gardner@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>
OSDs on PVCs previously were always replaced with PVCs of a given name.
For on-prem scenarios where the OSDs might need to be removed instead
of replaced, the operator now allows holes to exist in the PVC index
names as long as there are a sufficient number of PVCs to meet the
criteria for the deviceSet.count.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
Just like we do for the monitors, we set an ownerref to the OSD PVC so
when the cluster is deleted we remove the PVC too.
Signed-off-by: Sébastien Han <seb@redhat.com>
When the Ceph cluster runs on PVC and the OSDs are encrypted we can
store LUKS's Key Encryption Key inside a Key Management System. Today,
Rook only supports HashiCorp Vault: https://www.vaultproject.io/
The CephCluster has now a new "security" field which will plug onto the
KMS. Here is an example:
security:
kms:
tokenSecretName: <name of the secret containing a Vault token, used
to authenticate>
connectionDetails: < a map of strings containing connection
information>
Refer to the ceph-cluster-crd documentation to lear more.
Closes: https://github.com/rook/rook/issues/6105
Signed-off-by: Sébastien Han <seb@redhat.com>
on Azure, all SSDs are reporting as a rotational disk.
This commit will add `tuneFastDeviceClass` flag which will
tune for an SSD even if the cloud provider is reporting
rotational. We have a flag `tuneDeviceClass`
which works vice-versa.
Signed-off-by: subhamkrai <srai@redhat.com>
OSD prepare jobs only need to be run if the OSD deployments on PVCs were
not already started. For efficiency, we now query them in a batch instead
of querying them one-by-one with a label selector, which proves to be very
slow in large clusters.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
The OSD PVCs were being detected one by one with a label selector.
This proves to be slow in large clusters. Instead, we now make this
more efficient by querying for all existing PVCs and looking up their
existing in memory with a map.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
The pvc template name has either been require to be empty, or one of the
expected values such as data and metadata. Those who upgrade from 1.1
may find that if they changed it to something else, their upgrade will
fail when not one of these values. If there is only a single template,
always treat it as the data template.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
The operator will now allow the resource limits to be applied below the
recommended minimums. In small clusters, even the recommended minimums
may not be necessary. A warning is still printed to the operator log,
but we allow the configuration to continue.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
The osd prepare jobs and osd daemons may have different placement
requirements. In particular, the topology spread constraints
or pod anti-affinity should only work on either the prepare job or
osd daemon, but does not work effectively on both types at the same
time. Therefore, we split out the osd prepare placement to be separate
from the osd daemon placement. If the osd prepare placement is not set
the operator will fall back to the osd daemon placement for
backward compatibility.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
We can now encrypted OSD device that were provisioned via a storage
class using the PV interface.
The encryption works at the storageClassDeviceSets level, which means we
can have encrypted and non-encrypted sets.
Using the new key `encrypted` we can turn it on.
Signed-off-by: Sébastien Han <seb@redhat.com>
To provide more context for executing commands in a ceph cluster,
the full clusterInfo is now passed to the ceph execution commands.
All information about the cluster will now be available throughout
all the areas of the operator. The namespace, ceph credentials,
mon endpoints, and other info is a core part of that cluster info.
Arguments passed through the controllers are also simplified for
mons, mgr, osds, and other daemons where the parameters had
become too complex.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
The volume sources for an OSD on PVC were relying on the ordering
of the volume sources slice. While this was working, it was causing
headaches and potential future developer error when being
interpreted incorrectly. The addition of the wal pvc device
would make the issue even more pronounced. Now each PVC used
for the OSD will be identified in a map by name such as data
and metadata.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
When the PVC template increases in size for the storageClassDeviceSet
in the cluster CR, we need to increase the size of the PVC and restart
the OSD. The OSD is restarted by changing a variable on the pod spec
that indicates the desired size of the OSD.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
In 1.3 we added support for the metadata pvcs, thus requiring the name
to be set on the templates to differentiate between data and metadata.
For backward compatibility we need to assume the template is for data
if it is not set.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
Now, the CephBlockPool CRD is managed with the controller-runtime.
So the watcher is outside of the main controller reconciliation loop of
CephCluster which brings numerous benefit such as:
* having its own reconciliation loop
* won't block anything from the main CephCluster controller loop
* fast than waiting for CephCluster loop to completion
Partially close: https://github.com/rook/rook/issues/1981
Signed-off-by: Sébastien Han <seb@redhat.com>
The rook types used across the storage providers moved from the v1alpha2
package to the v1 package. This commit points the packages at their new
location. Implementation is expected to remain unchanged.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
When the OSD on PVC is backed by a metadata block PVC, the Ceph CRUSH
device class should be set to something else rather than the rotational
property of the drive.
Closes: https://github.com/rook/rook/issues/4881
Signed-off-by: Sébastien Han <seb@redhat.com>
We now support the addition of the PVC that acts as a metadata device
for a given OSD.
For this, you need to create a new `volumeClaimTemplates`, its name must
be "metadata" otherwise, Rook won't pick it up.
A template will look like this:
```
volumeClaimTemplates:
- metadata:
name: data
spec:
resources:
requests:
storage: 10Gi
# IMPORTANT: Change the storage class depending on your environment (e.g. local-storage, gp2)
storageClassName: gp2
volumeMode: Block
accessModes:
- ReadWriteOnce
- metadata:
name: metadata
spec:
resources:
requests:
storage: 6Gi
# IMPORTANT: Change the storage class depending on your environment (e.g. local-storage, gp2)
storageClassName: gp2
volumeMode: Block
accessModes:
- ReadWriteOnce
```
We now map block and block.db directly inside the container instead of
running ceph-volume activate. This is much cleaner.
Closes: https://github.com/rook/rook/issues/3852
Signed-off-by: Sébastien Han <seb@redhat.com>
TuneSlowDeviceClass wasn't part of the StorageClassDeviceSet types but
only in VolumeSource which is not exposed in the CR spec so setting the
option would never work.
The only issue is that this apply to all volumeClaimTemplates
where each can point to a particular storage class...
Signed-off-by: Sébastien Han <seb@redhat.com>
When using the "errors" package, using `%+v` (extended format),
each Frame of the error's StackTrace will be printed in detail.
Let's only print `%v` to print the error.
If the error has a Cause it will be printed recursively.
Basically `%+v` has been replaced with `%v` for all `error` type
interfaces, whether the logger is Info, Warning or Error.
Signed-off-by: Sébastien Han <seb@redhat.com>
OSD on PVC does not upgrade when the user upgrades the ceph version on cluster-on-pvc yaml.
The solution incudes:
- upgrade osd prepare and daemon pods on upgrade
- skip c-v prepare if filesystem is already present on the pvc device.
- skip lvm release in case of upgrade.
Signed-off-by: Santosh Pillai <sapillai@redhat.com>
Travis Nielsen
Carlos Barria
parth-gr
subhamkrai
Redouane KachachandBlaine Gardner
YZ775
parth-grandBlaine Gardner
Divyansh Kamboj
Yuichiro Ueno
Sébastien Han
Shachar Sharon
Nitin Goyal
Satoru Takeuchi
Arun Kumar Mohan
subhamkrai
UMEZAWA Takeshi
Santosh Pillai
rohan47