2b0c73a1c8
Improve the Ceph status check by calling the Ceph restful API using curl instead of the Ceph CLI. Uses the Ceph CLI as a fallback when the restful api is not working. The Ceph CLI takes around 250 ms to reply to a request, while the restful API takes around 35 ms for the same request. It reduces the load when checking Ceph status. To make rest API calls using curl, it needs the service url and the restful key. Since it takes time and resources to acquire those values, they are cached in the /var/run/ceph_hang directory. The .ceph_mgr_restful_key keeps the restful API key and the .ceph_mgr_service keeps the url to the service. Test-Plan: Test coverage: - Read/parse osd flag noup with error string injection; - Read/parse Ceph osd tree with error string injection; - Read/parse Ceph status with error string injection; - Read/parse Ceph health detail with error string injection; - Fallback to Ceph CLI when restful API is not available; - Swact controllers; - Power cycle standby controller; - Power cycle active controller; - Power cycle compute/storage node with ceph monitor; - DOR testing. PASS: AIO-SX PASS: AIO-DX PASS: Standard 2+2 PASS: Storage 2+2+2 Closes-bug: 2077673 Signed-off-by: Felipe Sanches Zanoni <Felipe.SanchesZanoni@windriver.com> Change-Id: I635bb0dc8dc674ff92567f71be6305271a6f87e0
integ
StarlingX Integration