The daemon name always needs to have the prefix rook-ceph-mds-
removed from the deployment name in order to translate to the
daemon name. The daemon name is not dependent on the namespace.
Signed-off-by: travisn <tnielsen@redhat.com>
The Nautilus v14.2.2 release was out last week. This is the new
default version for Rook in the documentation and manifests.
Signed-off-by: travisn <tnielsen@redhat.com>
The upgrade changes for #2901 added the check for the correct version
of the ceph image before continuing with an upgrade. This commit is
to refactor that change to work with the new code path to validate
the ceph version.
Signed-off-by: travisn <tnielsen@redhat.com>
If the cluster is not successfully initialized during the onAdd event,
the cluster will not be in a properly initialized state. The admin
will likely need to update the CR, in which case the onUpdate event would
be triggered. In this case we need to call the same code path as the
onAdd event to properly initialize the cluster.
Signed-off-by: travisn <tnielsen@redhat.com>
With the 1.1 release we should not be creating bluestore OSDs
with rook's legacy partitioning scheme. All new bluestore OSDs
should be created with ceph-volume. Therefore, the miminum
version allowed is a version of mimic supporting ceph-volume.
This commit also adds consistent version checking between
the Add and Update methods for the operator handling the
CR events.
Signed-off-by: travisn <tnielsen@redhat.com>
Range checks for CR values are done statically in the CRD
for the number of mons. There is no need to check in the code
if the mon count is out of range.
Signed-off-by: travisn <tnielsen@redhat.com>
This is not changing the behavior of the code but making the
flow of logic in the createReplicationCrushRule() function consistent
by removing an else branch. This was just a reflex when I was reading
the code.
Signed-off-by: Michael Adam <obnox@redhat.com>
Previously, if csi was not supported by the k8s version the csi set-up
code was skipped but csi was "left on". This change ensures that if
csi is not supported the csi enablement flags are set to false so
that other code that needs csi support will not run.
Signed-off-by: John Mulligan <jmulligan@redhat.com>
Return an error if the function that updates the csi config map
can not determine a valid namespace for the config map.
Signed-off-by: John Mulligan <jmulligan@redhat.com>
Add a prefix to the csi drivername when ceph csi is deployed by rook.
This aims to prevent trivial conflict between csi instances when two
rook operators are deployed into the same cluster. By default the
prefix is the "<rook-namespace>." but can by customized on the cli
and via environment vars.
Signed-off-by: John Mulligan <jmulligan@redhat.com>
Avoid using global values that are not used anywhere in the code.
Split the object used to express the ceph csi templates into two
so only values that are set globally need to be managed globally,
all other values can be stored in the temporary local struct.
Signed-off-by: John Mulligan <jmulligan@redhat.com>
Instead of requring csi instances for every ceph cluster that rook
manages, maintain one config map for all ceph clusters. This allows
for one set of csi drivers per rook operator.
The config is intialized as empty when Ceph CSI components are first
created and then updated every time a cluster's mons change.
As before clusters are identified by the k8s namespace the ceph cluster
is running in.
Signed-off-by: John Mulligan <jmulligan@redhat.com>
Create a mutex at the rook operator level that will be used when
updating the shared-between-clusters config map that Ceph CSI consumes.
Pass the mutex down to the function that will update the CM.
Signed-off-by: John Mulligan <jmulligan@redhat.com>
Thes values being defaulted "on" cause issues with some of the units
tests but add no additional value to the normal rook ceph binaries
because they are immediately overwritten by the values set on the
command line (and env).
Signed-off-by: John Mulligan <jmulligan@redhat.com>
Add a new function that, given existing config string, key, and current
mons, will return an updated version of the Ceph CSI config string
having updated the data with the new mons.
Add a unit test function for the update function.
Signed-off-by: John Mulligan <jmulligan@redhat.com>
- Added code to support StorageClassDeviceSet spec provided in the cluster-on-pvc.yaml
- The code reads the StorageClassDeviceSet spec and creates pvc based on the ‘count’ field for each device set.
- OSD prepare job is started for each PVC which activates the ceph-volume on each PVC
- Finally OSD is started on each of the PVC device.
Co-authored-by: rohan47 <rohgupta@redhat.com>
Co-authored-by: Ashish Ranjan <aranjan@redhat.com>
Signed-off-by: Santosh Pillai <sapillai@redhat.com>
when the root of a filesystem mode pvc is mounted as the base data directory
for the monitor it will not be empty: `lost+found` from ext4 will be present.
this causes the mon mkfs to fail because it expects an empty directory.
the solution is to mount under a directory in the source volume.
Signed-off-by: Noah Watkins <noahwatkins@gmail.com>
the life cycle post-start hook is used to chown the ceph daemon log
directory. however, hooks are not guaranteed to run before a container
entry point.
"Kubernetes sends the postStart event immediately after the Container is
created. There is no guarantee, however, that the postStart handler is
called before the Container’s entrypoint is called"
Signed-off-by: Noah Watkins <noahwatkins@gmail.com>
With this change the memory limit is only checked if it has been set
explicitely in the cluster CRD file.
This allow the pod to be created following the Burstable QoS strategy.
Before this change an error was raised (error trying to get the memory
limit) in the spec.CheckPodMemory function and the pod with only a 'memory request' set, was not able to start.
Resolves: #3564
Signed-off-by: Juan Miguel Olmo Martínez jolmomar@redhat.com
Signed-off-by: Juan Miguel Olmo Martínez <jolmomar@redhat.com>
The CSI driver requires the ceph key for administrating
the cluster. The key is stored in a secret and specified in the
storage class. The operator will now automatically generate
a secret named rook-ceph-csi that can be specified by
default in the storage class instead of requiring the
admin to create it.
Signed-off-by: travisn <tnielsen@redhat.com>
The discovery daemon is only necessary to be run when
OSDs are being created on raw devices and detect when
new devices are added to the cluster. If OSDs do not need
to be configured on devices, the discovery daemonset
has no need to be started.
Signed-off-by: travisn <tnielsen@redhat.com>
The flex driver does not need to be started if
only the CSI drivers are going to be used. Therefore,
we allow the admin to stop launching the rook flex
agent with the setting ROOK_ENABLE_FLEX_DRIVER in
operator.yaml
Signed-off-by: travisn <tnielsen@redhat.com>
CSI is now the preferred storage driver for Rook.
By default both the CSI and flex drivers will be started
by the operator. In a future release the flex driver
will be deprecated, but for now is still supported.
The drivers can be disabled with an environment
variable in the operator deployment in operator.yaml.
If users are only using one driver or the other there is
no need to enable both drivers.
Signed-off-by: travisn <tnielsen@redhat.com>
a helper flag indicating the exitence of an empty zone (empty = no
monitor pods) is computed when checking the health of the cluster in
order to determine if mons from overloaded zones have some new location
they can be scheduled. the computation of this flag did not check that
the empty zone contains valid nodes that pods can be scheduled on to.
this lead to a mon creation loop.
fixes: #3545
Signed-off-by: Noah Watkins <noahwatkins@gmail.com>
Replace usage of the custom osd copybins command with the new top-level
copy-binaries command introduced with cmd-reporter.
Signed-off-by: Blaine Gardner <blaine.gardner@suse.com>
In order to cleanup all cluster resources in case of cluster deletion
all resources have an ownerreference to the cluster crd instance.
Because the name of the reference was incorrectly set to the name of the
namespace the cluster resources can unintentionally be deleted by the
kubernetes garbage collector after a kubernetes cluster anomaly.
Signed-off-by: René Kalff <r.kalff@intermax.nl>
this adds the access for rook to create monitoring resources
across namespaces and provides servicemonitor config to
monitor the current service being created
Signed-off-by: Umanga Chapagain <chapagainumanga@gmail.com>
Travis Nielsen
Guangming Wang
Michael Adam
John Mulligan
rohan47andAshish Ranjan
Noah Watkins
Juan Miguel Olmo Martínez
Dmitry Yusupov
sabbot
Michael Vollman
AllenZMC
Blaine Gardner
René Kalff
Umanga Chapagain
Nick McKinney