v0.5.2

This release adds two additional flags for seastat server to control the maximum number of requests to Jolokia at any one time (concurrency) and maximum duration before a Jolokia request times out (timeout)

If you find Jolokia getting stuck and timing out, consider tuning down the concurrency parameter (which control
the maximum number of Jolokia requests in flight at any one time).

$ ./seastat server -p 8080 --concurrency 2

I've found 10 to be a good middleground for a Cassandra instance running on a beefy machine (8+ cores and 32GB of heap space) but your experience may vary.

We also expose a flag to configure the default Jolokia timeout per request (which you can use in combination with any other flags)

$ ./seastat server -p 8080 --timeout 2s
$ ./seastat server -p 8080 --concurrency=5 --timeout 2s

It is recommended to not set the timeout too high. A high timeout indicates Jolokia struggling to serve all the metrics needed.

v0.5.1

This adds a single metric to export the compression ration per table

Name	Description	Type
seastat_table_compression_ratio	Compression ratio for the table (a ratio of compressed size over uncompressed size)	Gauge

v0.5.0

v0.5.0 is out! 🍿

This release brings some improvements to the seastat_storage_node_status metric, specifically there is now a clear tag state distinguishing whether a node is up or down in the cluster.

This means we go from

seastat_storage_node_status{node="172.17.0.2",status="live"} 1

To this

seastat_storage_node_status{node="172.17.0.2",state="up",status="live"} 1

We also have two new metrics being exported

Name	Description	Type
seastat_internal_exceptions	Number of internal uncaught exceptions	Counter
seastat_storage_node_host_id	IP and Node UUID of each node that is part of the ring	Gauge

v0.4.0

v0.4.0 is out in the wild 🌤️

This release adds two new metrics for hints, thanks to @lyubent for the contribution!

Name	Description	Type
seastat_total_hints	Number of hint messages written to this node since [re]start. Includes one entry for each host to be hinted per hint.	Counter
seastat_total_hints_in_progress	Number of hints attempting to be sent currently from this node.	Gauge

v0.3.2

v0.3.2 is now out 🎉

(apologies, 0.3.0 and 0.3.1 used the wrong attribute name hence the metrics for compare and set operations weren't being scraped correctly 😓)

This release adds two table level metrics for Compare and Set operations

Name	Description	Type
seastat_table_cas_propose_latency_seconds	Compare and Set Propose Latency for queries	Summary
seastat_table_cas_commit_latency_seconds	Compare and Set Commit Latency for queries	Summary

v0.2.6

Fixes an issue where the Cells Scanned histogram was using the wrong underlying metric

v0.2.5 didn't seem to have the fix, this release fixes that

v0.2.4

Fixes a bug where floating point metrics were being read as integers. This may have reduced precision or (in the case of ratios), tended the metric to zero.

Affected metrics:

• TableStats: BloomFilterFalseRatio
• TableStats: KeyCacheHitRate
• TableStats: PercentRepaired

v0.2.3

This makes an optimisation to v0.2.1 to use the Bulk Request API for Storage Metrics. This will help significantly in the cases where you have a lot of keyspaces

Note that v0.2.2 had the same optimisation but was rendered less effective as it still grabbed everything.

v0.2.1

v0.2.0 (and v0.2.1) are now out 🎉

This release introduces Storage Metrics. These metrics come from Cassandra's storage service which keeps track of the cluster state from the perspective of each node

Name	Description	Type
seastat_storage_keyspaces	Number of keyspaces reported by Cassandra	Gauge
seastat_storage_tokens	Number of tokens reported by Cassandra	Gauge
seastat_storage_node_status	Status (live, unreachable, joining, moving, leaving) of each node in the cluster (tagged by node and status)	Gauge

v0.2.1 specifically Fixes a bug in seastat_storage_node_status where the value would not be 1.0 but rather the number of keyspaces you have in your cluster

v0.1.2

This is a bug fix release for some of the summary metrics added in v0.1.1. It fixes an issue where they weren't being parsed correctly and thus emitted zero values instead of the actual value