Spyglass on Streams

The 6.0 release of the MapR Converged Data Platform introduces Spyglass on Streams. When you install the the 6.0 version of MapR, Streams is the default mechanism through which metrics flow from the collectd service to OpenTSDB. Moving metrics through streams secures the data and provides a mechanism to perform real-time data analytics.

Note: Currently, Spyglass on Streams is not available for logs. Fluentd continues to use the REST API to send logs to ElasticSearch for the indexing of logs.

The Flow of Metrics via Streams

The collectd service collects node-level and service-level metrics from each node in the cluster. The collectd service hashes metrics to a stream and writes the metrics into topics in that stream.

Collectd creates two streams per cluster, Stream (0) and Stream (1). Each stream contains approximately 100+ topics. Topic names are of the format <hostname>_<metricname>, for example mfs81.qa.lab_cpu.percent.

Collectd maps the metric names to Stream (0) or Stream (1) using the djb2 hash algorithm, as shown:

static int hash(const char *str, int range)
{ int hash = 5381; int c; while ((c = *str++) != 0) hash = ((hash << 5) + hash) + c; /* hash * 33 + c */ return abs(hash%range); }

The algorithm hashes the metric name to an integer, either 0 or 1.

The Streams server distributes metrics through streams to the available OpenTSDB nodes and OpenTSDB consumes the metrics.
Note: Writing to an external OpenTSDB is not supported from release 6.0 onwards.

Spyglass on Streams Performance

The performance of Streams for metrics depends on the size of the cluster, the number of OpenTSDB nodes available to consume metrics data, and the number of streams and partitions available to move the metrics data from the collectd service to OpenTSDB.

Default collectd and OpenTSDB settings work well for clusters with up to one-hundred nodes and three OpenTSDB nodes. If the number of nodes increases or you notice that performance is sluggish, you can improve performance by adding OpenTSDB nodes and modifying the number of streams.

Evaluating Streams Performance

You can use the stream cursor list and stream topic info maprcli commands to view the producer (collectd) and consumer (OpenTSDB) statistics. Check the statistics to see if there is an increase in lag time between producers and consumers. If you notice an increase in lag time, increase the number of consumers (OpenTSDB nodes) or modify the streams and partition settings, as explained in the following sections.

For more information, see Monitoring Producers and Monitoring Consumers.

Determining How Many OpenTSDB Nodes to Install

Having multiple OpenTSDB nodes in the cluster distributes the workload. The number of partitions and OpenTSDB nodes determines the level of parallelism for consumption.

Each OpenTSDB node can consume one partition at a time. By default, metrics data is divided across five partitions in each topic and optimal parallelism is reached if there are five OpenTSDB nodes to consume the partitions. See Parallelism When Consuming Messages and note that the term “consumer” in the topic equates to an OpenTSDB node in Spyglass on Streams.

A general guideline for the minimum number of OpenTSDB nodes in a cluster is one for every 10x increase in nodes beyond ten, for example:
  • 3 OpenTSDB nodes in a 10 node cluster
  • 4 OpenTSDB nodes in a 100 node cluster
  • 5 OpenTSDB nodes in a 1000 node cluster
  • ...
Note: For every 10x nodes you install, you should increase the streams number by 2.

If your cluster has ten or more nodes, at least three OpenTSDB nodes should be available to consume metrics. Typically, for every 10x increase in nodes, you should add another OpenTSDB node. For example, if your cluster reaches a size of 100 nodes, have 4 OpenTSDB nodes available for consumption. Note that an increase in the number of OpenTSDB nodes may require an increase in the number of streams and/or partitions.

These guidelines do not guarantee optimal performance. Evaluate the performance of the streams to determine if your cluster would benefit from additional OpenTSDB nodes or tuning the number of streams or partitions.
Note: If all configured OpenTSDB nodes have been offline for several hours, you may notice an initial spike in memory and CPU usage by OpenTSDB processes as they aggressively try to keep up with the metrics. You can reduce the number of AsynchHBase threads to reduce the CPU and memory usage. By default, AsynchHBase runs 128 threads. To modify the number of threads, add/modify the following property in the /opt/mapr/asynchbase/asynchbase-<version>/conf/asynchbase.conf file on the OpenTSDB nodes:

Increasing the Number of Streams

The default setting for the number of streams is two. As a general guideline, for every 10x increase in the number of cluster nodes, add two additional streams. For example, if your cluster has one-hundred nodes, add two more streams, for a total of four.

To increase the number of streams, edit the streams-specific options in the collectd and OpenTSDB configuration files. The streams option in each file must have the same value. After you change streams parameters, reconfigure MapR Monitoring, as shown in Update the Monitoring Storage Nodes.

The following table lists the streams parameters, the default setting, suggested setting for 100 and 1000 nodes, and the file in which each parameter is located:
Parameter Default Setting (up to 100 nodes) Number of Streams for 100 nodes Number of Streams for 1000 nodes File Location
StreamsCount 2 4 6 /opt/mapr/collectd/collectd-<version>/etc/collectd.conf
MAXSTREAMS 2 4 6 /opt/mapr/collectd/collectd-<version>/etc/init.d/collectd
tsd.streams.count 2 4 6 /opt/mapr/opentsdb/opentsdb-<version>/etc/opentsdb/opentsdb.conf
Note: The settings in the table are suggestions based solely on the number of nodes in the cluster. The suggested settings do not guarantee optimal performance.

Changing the Automatic Stream Cursor Commits

You can adjust the frequency of automatic stream cursor commits for OpenTSDB. Modify the tsd.streams.autocommit.interval in opentsdb.conf The unit is thousands of seconds. The default value is '60000' which is 60 secs. For a system with heavy loads, consider changing the value to something like 5 minutes.