Added a provisioner toleration to the csi version
check job, so that we can schedule the csi version
check on the node where the nodes are tainted.
Signed-off-by: Madhu Rajanna <madhupr007@gmail.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>
Somehow, the client.Update was not updating the cluster object, using
client.Status().Update actually fixed it.
Signed-off-by: Sébastien Han <seb@redhat.com>
Instead of setting the Phase of the CR once the reconcile is done, let's
actually set in from the healthcheck so it is more accurate.
Closes: https://github.com/rook/rook/issues/5249
Signed-off-by: Sébastien Han <seb@redhat.com>
When we run the goroutine, we should run start it immediately instead of
waiting for the configured interval to start. With that we can get the
Status Phase reported much faster in the CRD.
Signed-off-by: Sébastien Han <seb@redhat.com>
At startup there is a race condition between the goroutine checking
for mon health and the first reconcile of the mons. There is a lock
that ensures only one of them will check the mon health at a time.
The issue is that if the wrong one gets there first, it will not
have the desired mon count initialized from the cluster CR as expected
and the healther checker will use a desired mon count of 0.
The desired count of 0 leads the health checker to remove
a mon since there are extra mons. Only one mon is allowed
to be removed at a time, and adjusting to fewer than two mons is
disallowed once there are at least two mons.
After the health checker is completed, the main reconcile of the mons
will proceed with the correct desired mon count of 3 and attempt
to restore to the real desired quorum size.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
The mon was starting with a partial spec. There was a race condition
that was causing incorrect initialization in the health check goroutine
that could remove a mon upon operator restart. Now the mon will be
properly initialized with the full spec.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
The admin needs to start the toolbox and run a command such as ceph osd tree to see the device classes for each OSD. This PR simplifies it by storing the deviceclass info in the CephCluster CR status
closes: https://github.com/rook/rook/issues/5754
Signed-off-by: kesavan <kvellalo@redhat.com>
- Add realm/zone group/zone to Object Context,
so that any call to runAdminCommand has the
same realm, zone group, and zone as the object
store that the call is going to.
- Also modify the delete code for the object-store
using multisite to remove the endpoints of an
object store from a zone instead of the normal
clean-up
- Added more debug logging all around the object-store
code related to multisite
- files generated by rerun of `make codegen`
- change back the edit on the Copyright in object.go
Signed-off-by: Ali Maredia <amaredia@redhat.com>
When zones are created the ceph RGWs inside the
those zones should be using pools with the zones
name not the object-store's name
Signed-off-by: Ali Maredia <amaredia@redhat.com>
This commit:
- adds the pull section on the CephObjectRealm spec
to enables realms to be pulled instead of created.
- creates the system user and generates the access
key and secret key for the user and zones in a realm.
- adds "omitempty" to fields in multisite CRs that
are not required
Signed-off-by: Ali Maredia <amaredia@redhat.com>
During the deletion sequence of the CR, we now set a success flag at the
end, right before removing the finalizer.
This avoids the case where we fail to remove the finalizer.
2020-07-21 13:29:46.334816 E | ceph-object-controller: failed to reconcile failed to remove finalizer: failed to remove finalizer "cephobjectstore.ceph.rook.io" on "teststore": Operation cannot be fulfilled on cephobjectstores.ceph.rook.io "teststore": the object has been modified; please apply your changes to the latest version and try again
Signed-off-by: Sébastien Han <seb@redhat.com>
Now, we can not only cleanup monitor data, logs and crashes but the
disks too. As part of the cleanupPolicy CR spec, we have a new setting
called sanitizeDisk which holds more details:
* method: indicates if the entire disk should be sanitized or simply ceph's metadata.
Possible choices are 'complete' or 'quick' (default)
* dataSource: indicate where to get random bytes from to write on the disk.
Possible choices are 'zero' (default) or 'random'
Using random sources will consume entropy from the system and will take much more time then the zero source
* iteration: overwrite N times instead of the default (1). Takes an integer value
See the documentation for more details.
Signed-off-by: Sébastien Han <seb@redhat.com>
The label generated to apply to a node are simplified with the
backtick string delimeter to avoid the escaping of quotes.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
The ClusterInfo does not use the Log method to write the fields
anymore. The formatting was not necessary since the struct can
be logged if needed with the standard %v golang formatting.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
All assignments made to errors or other variables should be
used instead of ignored. This checks for errors in a couple
places that were previously ignored, and also initializes
a variable such that it won't be always overwritten by the
various if else statements.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
The user creds should not be passed to the ceph helper methods
since the clusterInfo already has the needed context. Removing
a final reference that was missed in the refactor.
Signed-off-by: Travis Nielsen <tnielsen@redhat.com>
ceph-file-controller gets stuck in loops, this happens when it tires to delete the filesystem which doesn't exist.Filesystem should only be deleted when it exist.
Signed-off-by: RAJAT SINGH <rajasing@redhat.com>
If the healthcheck goroutine cannot check for the ceph status/times out
(if the mons are not in quorum) this should still be reflected in the
CRD status field.
Example:
```
status:
ceph:
details:
error:
message: 'failed to get status. . timed out: exit status 1'
severity: Urgent
health: HEALTH_ERR
lastChanged: "2020-07-20T14:11:05Z"
lastChecked: "2020-07-20T14:11:05Z"
previousHealth: HEALTH_WARN
```
Signed-off-by: Sébastien Han <seb@redhat.com>
A long-standing issue that was only solved via a documentation note. Now,
during the prepare pod instantiation, we check for the presence of the
lvm binary on the host. Success will indicate that we can bootstrap
OSDs where failure will refuse to prepare the OSD.
Closes: https://github.com/rook/rook/issues/5627
Signed-off-by: Sébastien Han <seb@redhat.com>
Name for requeue time variables were the same for
the object-zone and the object-zonegroup.
This commit makes them more descriptive and
accurate
Signed-off-by: Ali Maredia <amaredia@redhat.com>
The masterArg flag cannot be "" so a different
command without masterArg must be run for zone
and zone group creation
Signed-off-by: Ali Maredia <amaredia@redhat.com>
Make it so that zones & zone groups can be
modified with more than one endpoint so that
more than one object store can be in a zone
Signed-off-by: Ali Maredia <amaredia@redhat.com>
Pass desired devices to the OSD provisioning container by
JSON-marshalling/-unmarshalling a disk ID with StorageConfig settings.
This will allow disks to be specified that contain special characters.
Notably, this will support /dev/disk/by-path/pci-HHHH:HH:HH.H devices
that were unsupported previously due to the format which separated
devices from params with colons (:).
Fixes#5056Fixes#5535
Signed-off-by: Blaine Gardner <blaine.gardner@suse.com>
Travis Nielsen
subhamkrai
Sébastien Han
Madhu Rajanna
kesavan
Ali Maredia
Hiroshi Muraoka
RAJAT SINGH
Blaine Gardner