- Operator is deployed under namespace rook-system
- Operator deploys rook-agent on all nodes as daemonset also on
rook-system
- Rook-agent installs flexvolume driver on hosts
- Rook-agent listens on unix socket for driver requests
- Rook-agent perform attach/detach on its node
- Rook-agent creates/delete CRD volumeattach objects
- Added fencing to support ROX and RWO
- Added unit and integration tests
- Updated examples and docks
fixes#432
- multiple file systems are allowed
- multiple mds instances are allowed per file system.
- a keyring is created per mds instance
- double the desired mds instances are started, half in standby mode
separately (240s)
Added comments
operator/mon: made the doc for the nodeState better
Renamed the monHealth to monTimeoutList
Renamed the monOutInterval to monOutTimeout
operator/mon: Implemented the feedback
tests/smoke: Corrected mon_test comment
smoke/mon: set interval and timeout for health check
Added build constraint to smoke/mon_test for linux
As written here https://github.com/rook/rook/pull/958#issuecomment-327507930
Removed controller.go healthCheckInterval
Set the interval for mon_test in smoke_test SetupSuite()
Fixed the smoke test by adding the build constraint
Removed obsolete clientTimeout from operator/cluster
Ran gofmt on block_test and helm_test
Use flags for the mon healthcheck and mon out timeout config
Added helm chart value sfor healthCheckInterval and monOutTimeout
Fixed the timeout map in mon/health.go
Fixed issues with some tests stuff
Reference the operator/mon defaults in the flags
Added Documentation for Ceph mon failover times env vars
Added missing toc link in documentation
Removed "Tweak Ceph mon failover times" to comments in rook-operator.yml
First, this depends on the MIN_SIZE parameter of a pool. Second even if the
MIN_SIZE is one this would only hold up to six node cluster. Imagine a nine
node cluster; losing two-thirds of a cluster can lead to a situation where all
replicas of an object are offline. I.e., you can only lose up to two failure
domains, in this case nodes, when you have a replication size of three.
refactor integration tests splitting out the kubernetes
cluster setup to just be in a script and making it the reponsibility
of the host/user/jenkins.
integration tests now run just like unit tests using "go test".
Gerard Hickey
Travis Nielsen
Alexander Trost
Jared Watts
Pavel Snagovsky
Steve Leon
Gerard Hickey
Hans van den Bogert
Mahmoud Reza Rahbar Azad
Bassam Tabbara
Ian Lovett
Sergi Almacellas Abellana
talkdirty
bzub