4b7e7bf905
Cassandra Time to Live (TTL) is applied per-column in a row, and with a default TTL for the table. The Cassandra driver appears to have been written with a misunderstanding that the TTL was per-row, and as a result the update statements were written to avoid writing all cells. But that results in some cells in the row not being updated, and after their TTL expires they are removed, so a metric query may result in partial data (updated_at set but metric_id and created_at missing). This fix changes the statement used when a new measurement is received to use the same insert statement for the metric row and thus refresh the TTL for all the columns. A follow on patch should be created to remove the Update statements for Measurements, and consider refactoring/removing the caching model. Change-Id: I6b7636a52e8bdb2ce8ad97f839acb6184cd58a8b Story: 2005832 Task: 35872 |
||
|---|---|---|
| common | ||
| config-generator | ||
| docker | ||
| etc/monasca | ||
| java | ||
| monasca_persister | ||
| perf | ||
| playbooks | ||
| releasenotes | ||
| .coveragerc | ||
| .gitignore | ||
| .gitreview | ||
| .stestr.conf | ||
| .testr.conf | ||
| .zuul.yaml | ||
| bindep.txt | ||
| LICENSE | ||
| lower-constraints.txt | ||
| pom.xml | ||
| README.rst | ||
| requirements.txt | ||
| run_maven.sh | ||
| setup.cfg | ||
| setup.py | ||
| test-requirements.txt | ||
| tox.ini | ||
Team and repository tags
monasca-persister
The Monitoring Persister consumes metrics and alarm state transitions from the Message Queue and stores them in the Metrics and Alarms database.
Although the Persister isn't primarily a Web service it uses DropWizard, https://dropwizard.github.io/dropwizard/, which provides a nice Web application framework to expose an http endpoint that provides an interface through which metrics about the Persister can be queried as well as health status.
The basic design of the Persister is to have one Kafka consumer publish to a Disruptor, https://github.com/LMAX-Exchange/disruptor, that has output processors. The output processors use prepared batch statements to write to the Metrics and Alarms database.
The number of output processors/threads in the Persister can be specified to scale to more messages. To horizontally scale and provide fault-tolerance any number of Persisters can be started as consumers from the Message Queue.
Build
Requires monasca-common from https://github.com/openstack/monasca-common. Download and build following instructions in its README.md. Then build monasca-persister by:
mvn clean packageConfiguration
A sample configuration file is available in java/src/deb/etc/persister-config.yml-sample.
A second configuration file is provided in java/src/main/resources/persister-config.yml for use with the vagrant "mini-mon" development environment.
TODO
- Purge metrics on shutdown
- Add more robust offset management in Kafka. Currently, the offset is advanced as each message is read. If the Persister stops after the metric has been read and prior to it being committed to the Metrics and Alarms database, the metric will be lost.
- Add better handling of SQL exceptions.
- Complete health check.
- Specify and document the names of the metrics that are available for monitoring of the Persister.
- Document the yaml configuration parameters.
License
Copyright (c) 2014 Hewlett-Packard Development Company, L.P.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.
Python
To install the Python monasca-persister modules, git clone the source and run the following command:
$ sudo python setup.py installTo run the unit tests use:
$ tox -e py27,py35