ceph: refactor mgr-restful-plugin state machine

Browse Source

Replace mgr-restful-plugin service monitoring based on state
machine with explicit transitions with a function that goes
through the following steps:
- wait for Ceph cluster to become available
- configure and start ceph-mgr
- configure and enable restful plugin
- send periodic requests to REST API

Procedure to recover from errors: restart ceph-mgr, update
certificates, run again through configuration steps, wait
for Ceph cluster.

mgr-restful-plugin components:

1. init script: parse command line parameters, start service
   monitoring if not already running, request status via
   control socket, stop service monitoring.

2. service monitoring: create process running ceph-mgr
   confguration steps, report status based on current
   rest api availability and init/recovery procedure,
   stop ceph-mgr and helper process

3. monitoring helper: successively run commands to configure
   and start ceph-mgr and restful plugin, send periodic requests
   to ceph REST API, update service monitoring with current
   failure status (no response from REST API or ceph-mgr
   restarts).

When ceph-mgr fails too many times mgr-restful-plugin exists
and relies on SM to restart it (ceph-mgr is also restarted).

Change-Id: Id5342624948024ce2891e32ee6648c910a6e7391
Closes-Bug: 1828024
Signed-off-by: Daniel Badea <daniel.badea@windriver.com>

...

This commit is contained in:

Daniel Badea

2019-05-16 12:21:56 +00:00

parent ce0cc60346

commit a4759b8e5b

1 changed files with 821 additions and 518 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download Patch File Download Diff File Expand all files Collapse all files

1339

ceph/ceph/files/mgr-restful-plugin.py

File diff suppressed because it is too large Load Diff