retrieveMultisiteZone is meant to gate the object-store reconcile on the
backing Ceph zone existing: it runs "radosgw-admin zone get" and, when
that fails, returns a non-nil error so the caller requeues. That gate is
dead code. The ENOENT check declares an inner err from exec.ExtractExitCode
that shadows the outer command error, and ExtractExitCode returns a nil
error for the ordinary exit failures radosgw-admin produces. Both the
ENOENT branch and the else branch then wrap that shadowed nil, and
errors.Wrapf(nil, ...) is nil, so the function returns
(waitForRequeueIfObjectStoreNotReady, nil). The caller only propagates the
requeue when the error is non-nil, so the requeue is dropped and reconcile
runs on.
The result is that a failed "zone get" no longer backs off. Reconcile
proceeds to stand up the object store anyway -- the RGW service, the
admin-ops endpoint, the deployment, and the pools radosgw scaffolds as it
comes up -- for a multisite store whose backing zone does not exist. This
is not the "normal multisite bootstrap" transient the original wording
suggested. getMultisiteResourceNames runs immediately before this and
already requeues until the CephObjectZone CR reports Ready, and the zone
controller marks it Ready only after it has created the Ceph zone, so a
healthy bootstrap never reaches this gate with a missing zone. That
CR-Ready gate, not this one, is what actually blocks bootstrap.
Where the dead gate does bite is the cases the CR status cannot cover:
- the Ceph zone deleted or renamed out of band while the CR still reads Ready
- a zone controller that reports Ready without leaving a usable zone behind
- any non-ENOENT "zone get" failure, e.g. a permission or connectivity
error, which the else branch swallows the same way
The check was correct when fc579f4520 introduced it in 2020 with
exec.ExitStatus, which returns (code, ok) and leaves err unshadowed.
bd58790c31 ("ceph: proxy ceph commands when multus is configured", 2021)
swapped it to exec.ExtractExitCode as an unrelated drive-by, inverting the
contract and killing the gate. The sibling realm, zonegroup, and zone
controllers were left untouched and still use exec.ExitStatus today.
Restore exec.ExitStatus so the outer error is no longer shadowed, both
branches wrap the real command error, and the caller requeues until the
zone exists -- matching the sibling controllers.
Signed-off-by: Anas Khan <83116240+anxkhn@users.noreply.github.com>
Signed-off-by: Joshua Hoblitt <josh@hoblitt.com>
Fix duplicate words, incorrect articles (a/an), it's/its, and other small
grammar mistakes in Go comments and user-facing messages across pkg/, cmd/,
and tests/.
Signed-off-by: Joshua Hoblitt <josh@hoblitt.com>
with controller runtime latest version vO.23.0, golangci lint
is complaining about deprecated api deprecated.
This commit updates the api and adds the requried rbacs
Signed-off-by: subhamkrai <srai@redhat.com>
For all of the controllers besides the cluster controller,
the logging now includes the namespaced name of the resource
that is being reconciled. This will help with log troubleshooting
to help analyze logs consistently for the resource being
reconciled.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
The logger names are seen on each line of logging. As
long as they are unique, they really don't need to be
long and descriptive, so let's make them a bit more
concise, and independent from controller names.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
add defaultRealm field to cephobjectstore to allow rook to mark the
created realm as default. this prevents ceph from auto-creating a
"default" zonegroup and zone when the default is not explicitly set.
affects only non-multisite setups.
Signed-off-by: Oded Viner <oviner@redhat.com>
add RecoverAndLogException() helper to log panics with stack trace.
added defer call in all rook controller Reconcile() methods for
better error visibility in operator logs
Signed-off-by: Oded Viner <oviner@redhat.com>
This is the first implementation of CephX key rotation in Rook.
Adds key rotation API from design PR 15915.
Also add helper methods for rotating keys, determining when keys
need to be rotated, and for generating CephX key statuses. These will be
reusable for other Rook reconciles beyond object/RGW.
Signed-off-by: Blaine Gardner <blaine.gardner@ibm.com>
The object store controller was swallowing errors of the
type not found, assuming a certain category of errors
when ceph may still be initializing. But some of these
errors could be user error and they need to troubleshoot
the details of the error, so we need to log the specific
error instead of swallowing it.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
When finalizers are added to the CRs, a follow-up reconcile
will be triggered due to the increased generation on the CR.
Therefore, abort the initial reconcile when adding the finalizer,
and allow the follow-up reconcile to complete the configuration.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
All existing controller runtime watches are converted to use "typed"
handlers and predicates instead of operating on `client.Object`. The
intent is to be bug for bug equivalent with the existing logic while
replacing run time type assertions and switch statements with compile
time type constraints and type casts. In several cases, functions using
assertions were split up such that each function only handles a single
Kind at a time. It is hoped that this will improve readability and
maintainability while facilitating future refactoring such as migrating
some watches to using IndexFields.
Of particular note is that the massive switch statement in
`WatchControllerPredicate()` from
`pkg/operator/ceph/controller/predicate.go` has been replaced with
generics, reflection, and splitting the obc logic into its own predicate
function. There are still many helper functions operating on
`client.Object`. These were not updated unless required by the compiler
in order to limit the size of this change. The type safety of these
funcs should be improved as followup work.
It is strongly suggested that going forward, handlers and predicates
only handle a single Kind (generic or not) and that switches / type
assertions are heavily discouraged or forbidden. This PR removed all
but a single switch statement in a predicate, which should be addressed
in future work.
Signed-off-by: Joshua Hoblitt <josh@hoblitt.com>
Implements #15460. If ObjectStore reading value from a secret
provided by user, the secret will be annotated with ObjectStore
name and ObjectStore will be reconciled on secret change.
Signed-off-by: Artem Torubarov <artem.torubarov@clyso.com>
This reverts commit a941b3c33f.
Stop creating the 'cosi' user in the CephObjectStore reconcile. This
step often fails for some amount of time during initial object store
creation, causing frequent user concern. It has also been the source of
some reported failures that would otherwise be non-breaking for certain
users.
Signed-off-by: Blaine Gardner <blaine.gardner@ibm.com>
For the specification see:
<https://github.com/rook/rook/blob/master/design/ceph/object/swift-and-keystone-integration.md>
* extend the API object specs for swift and keystone integration
* adapt rgw to the new go-ceph version
- The parameter lists of the API call have changes, as parameters
ignored by the RGW Admin Ops API are no longer serialized, therefore
the mock has to be adapted.
- There is now validation for the user keys that are passed to the
User get API, therefore things failed when we had empty keys in our
User proxy object.
* expand the reconcile loop for the swift and keystone integration
* fix minor mistakes in design document
* add env var to pass extra args to minikube
Minikube decides CPU cores and memory automatically based on the
available resources on the machine which may be insufficient to
run rook. This commit adds an environment variable to add arbitrary
arguments to the minikube command, so both can be specified if
desired.
* integration tests for swift and keystone
The new integration of swift or s3 and keystone support by rook
does not have any integration tests yet.
This commit introduces integration tests for swift and keystone. The
tests are done against a minimal keystone setup (keystone container
image from Yaook-project (https://yaook.cloud), sqlite as database
backend, cert-manager and trust-manager for test certificate setup).
To prevent hardcoded credentials, passwords are generated
by the tests. The integration tests use the openstack client
(keystone- and swift-functionality) (https://docs.openstack.org/
python-openstackclient/ latest/). This was a concious design decision
to use client tooling as close as possible to the end user instead of
using other go-libraries (such as gophercloud).
* add documentation on swift and keystone
Currently there is no documentation on the use of Swift to access
an object store as well as the use of OpenStack keystone for
authentication.
This commit adds documentation on the use of Swift and OpenStack
keystone, as well as CRD-related documentation and an example setup.
* add integration tests for S3 via keystone
This commit introduces integration tests for s3 and keystone. The
tests are run against the same minimal keystone setup that the tests
for swift and keystone use.
The integration tests use the aws s3 client to use client tooling as
close as possible to the end user instead of using other go-libraries.
Co-authored-by: Jan Klippel <jan.klippel@uhurutec.com>
Co-authored-by: Silvio Ankermann <silvio.ankermann@cloudandheat.com>
Signed-off-by: Sebastian Riese <sebastian.riese@cloudandheat.com>
Signed-off-by: Jan Klippel <jan.klippel@uhurutec.com>
Signed-off-by: Silvio Ankermann <silvio.ankermann@cloudandheat.com>
Add CephObjectStore spec.hosting.advertiseEndpoint configuration. This
provides a clear documented default for which endpoint Rook "advertises"
to dependent resources like CephObjectStores, OBCs, and COSI
Buckets/Accesses and allows users to override the default behavior if
desired.
The current default is to round-robin an endpoint from
spec.hosting.dnsNames, which has proven to be troublesome for some
users' object store configurations. This change provides much-needed
disambiguation for users.
This may be a breaking change for some existing spec.hosting.dnsNames
users. This is unexpected but is documented.
Signed-off-by: Blaine Gardner <blaine.gardner@ibm.com>
Until now, an object store would create all the necessary
metadata pools and the data pool that were exclusively
for its own object store. When isolation between object
stores is necessary, this would cause many pools and
PGs to be created in the cluster, which was not
manageable.
Now one set of pools can be created to be shared
by any number of object stores. The metadata and data
between each object store is isolated by
RADOS namespaces, which by design will keep the
data safe for multi-tenancy.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
The bucket health checker was removed in 1.10. Now in 1.12
we no longer need this removal of the bucket health
checker since it will no longer exist to remove.
Signed-off-by: travisn <tnielsen@redhat.com>
The radosgw-admin command uses the network spec from ceph cluster spec
in object context but it is not filled properly in the object package.
But with PR 10898, network spec is available in clusterinfo which can
be used directly. Also removed cluserspec from object context.
Signed-off-by: Jiffin Tony Thottan <thottanjiffin@gmail.com>
External CephObjectStores already have endpoints defined by
spec.gateway.externalRgwEndpoints, and if the external store is
configured with TLS (HTTPS), the store's certificates will likely not
accept connections intended for the Service endpoint Rook creates. Some
users might not be able to easily add the service endpoint to their
certificates. Therefore, don't even bother creating a Service for
external clusters.
This does introduce a few issues. The Service seems to have been
initially created to allow multiple external RGW endpoints to be
addressable via a single address in Rook. For all connections to an
external CephObjectStore with multiple endpoints, simply choose an
endpoint at random. Random selection will prevent Rook from failing to
create buckets or users on an external store if one of the external
store's endpoints fails.
The latest OBC library (lib-bucket-provisioner) allows updating the
endpoints on ObjectBuckets after they are created. This allows Rook
users to change endpoints on external CephObjectStores without breaking
all existing OBCs. It requires implementation of the new GetUserID()
library call, requires updating Provision() and Grant() calls to be
idempotent, and it requires removing the Update() call.
Signed-off-by: Blaine Gardner <blaine.gardner@redhat.com>
Remove the health checker for CephObjectStore. The liveness and
readiness probes go through the same code paths in RGW as creating
buckets without as much affect on the storage backend.
Full discussion: https://github.com/rook/rook/issues/11031
Signed-off-by: Blaine Gardner <blaine.gardner@redhat.com>
if Multus is enabled the clusterinfo should be updated with
network as multus as to run the ceph cmds in remote
executor
Signed-off-by: parth-gr <paarora@redhat.com>
User can define his desired endpoint list in Zone CR so that it will
overwrite the default service name for rgw.
Resolves#6432
Signed-off-by: Jiffin Tony Thottan <thottanjiffin@gmail.com>
Signed-off-by: Jiffin Tony Thottan <jthottan@redhat.com>
Check the radosgw-admin realm user list per object store instead of relying
on the ceph dashboard get-rgw-api- command.
Resolves#9099
Signed-off-by: Alexander Trost <galexrt@googlemail.com>
The CSI package needs to load clusterInfo, today this code is in the mon
package which makes the call of LoadClusterInfo impossible without
having a circular import.
Signed-off-by: Sébastien Han <seb@redhat.com>
Previously, the struct maintaining the list of cluster was still
initialized with a cluster item. Then the monitoring check will see that
the cluster is part of the struct already and thus won't run the
monitoring go routine again.
Now each time we cancel the context, we also remove the cluster item
from the map so that when the controller runs again, the monitoring
struct is re-populated and the go routine runs and statuses are updated.
Closes: https://github.com/rook/rook/issues/9911
Signed-off-by: Sébastien Han <seb@redhat.com>
adding observedGeneration field in the cephcluster cr
status for having better control on reconciling,
as observedGeneration field will be updated by the controller
Closes: https://github.com/rook/rook/issues/9673
Signed-off-by: parth-gr <paarora@redhat.com>
ReportReconcileResult should never be given an object that is nil. It is
impossible to force compile-time checking for this because the object's
type is an interface. We can't force this, but we can rework the
function definition to handle more cases where the object may not be
returned fully-complete, and we can rework callers to return structs
(not pointers-to-structs) so it is less likely a caller will pass nil
to the function.
Signed-off-by: Blaine Gardner <blaine.gardner@redhat.com>
We need to explicitly assign the ceph version within each scope since
one function returns a pointer and one is not.
Previously, the `desiredCephVersion` was recreated in the scope of the
`else` and thus a new variable `desiredCephVersion` was initialized (see
the `:=`).
Now, each scope creates and assign to clusterInfo a `desiredCephVersion`
value.
Signed-off-by: Sébastien Han <seb@redhat.com>
The object child controller should not compare versions if the object
store is external. The current version comparison is using the
cmdreporter to check the ceph version of the image in the CephCluster spec.
In external mode, this image is not set so the cmdreporter fails.
Also, this check is only valid for converged mode, so external mode
should be skipped.
Signed-off-by: Sébastien Han <seb@redhat.com>
Prior to this, we were comparing a pointer (the memory address) with a
struct. This was obviously always failing and returned false. We must
dereference the pointer to access the data contained at that memory
location.
Closes: https://github.com/rook/rook/issues/9544
Signed-off-by: Sébastien Han <seb@redhat.com>
The original PR which added event reporting unnecessarily "optimized" to
prevent spamming the API controller[1].
It is sometimes important to get events as they happen and not hide new
events behind preexisting older events. For example, in integration
tests, we may often want to wait for a controller to finish processing
an update, and the best way to do that is to wait for the
"ReconcileSucceeded" event. In order for this to be useful, the events
must be reported each time.
If we begin having problems with events being reported too often, then
we should fix the underlying issue of reconciles happening too often
instead of relying on a time-based "optimization" that hides recent
event reports that may be useful.
[1]: https://github.com/rook/rook/pull/7222
Signed-off-by: Blaine Gardner <blaine.gardner@redhat.com>
Anas Khan
Joshua Hoblitt
Artem Torubarov
subhamkrai
Travis Nielsen
SantoshPillai
cuiweixie
Oded Viner
parth-gr
Blaine Gardner
Artem Torubarov
Artem Torubarov
Jiffin Tony Thottan
sp98
Blaine Gardner
parth-gr
Joseph Lee
Josh Soref
Alexander Trost
Sébastien Han
Divyansh Kamboj