monitor pending compaction tasks on the database node

[Raj Reddy]

we should collect some stats from nodetool such as compaction tasks, tpstats etc and send it from nodemgr
active,pending, all time blocked
Native transport, Mutation stage, Read stage, MemtableFlushWriter

[OpenContrail Admin (ci-admin-f)]

Review in progress for https://review.opencontrail.org/20681
Submitter: Arvind (<email address hidden>)

[OpenContrail Admin (ci-admin-f)]

Review in progress for https://review.opencontrail.org/20999
Submitter: Raj Reddy (<email address hidden>)

[OpenContrail Admin (ci-admin-f)]

Reviewed: https://review.opencontrail.org/20999
Committed: http://github.org/Juniper/contrail-controller/commit/214d61eb4e4eda19ba3bbd336e6f3c801c6f4e5d
Submitter: Zuul
Branch: master

commit 214d61eb4e4eda19ba3bbd336e6f3c801c6f4e5d
Author: arvindvis <email address hidden>
Date: Thu May 26 18:17:08 2016 -0700

This commit adds reporting of cassandra thread pool stats
for monitoring. It collects the nodestatus output and
reports it every min.
Closes-Bug: 1583733,1588156,1589039

(cherry picked from commit eda32030333f80247e3195d23aeb8449325d6658)

Conflicts:
 src/nodemgr/database_nodemgr/database_event_manager.py

send_database_status uses the inner structure
cassandra_compaction_task without initializing,
the fix takes care of that.

(cherry picked from commit 49f6637149803e8020084dff9b47359eb0e6f6f0)

The output of 'nodetool compactiontasks' can contain multiple lines
and hence we should grep for line having 'pending tasks' before
getting the value..

(cherry picked from commit f652617414b37534f9329bf0e0c94ad37262d6af)

Change-Id: I78993df71a32295e9697768e4b64592e6c4c9405