| 741ae0b | brharrington | 25 July 2018, 15:02:38 UTC | add timeRange helper function (#859) There are a number of internal libs that have this same logic of computing a time range based on parsed start and end times that can be relative to each other. Move to a helper function of Strings so it is easier to reuse. | 25 July 2018, 15:02:38 UTC |
| 28a40ce | brharrington | 24 July 2018, 18:18:11 UTC | short circuit pattern matching with prefix check (#858) In the streaming execution, up to 25% of the CPU usage is checking regular expressions. This change improves the chances that checks will be able to quickly fail without needing regex by always checking the fixed prefix when known in the matches function. This wasn't done before because when evaluating with the index we already know the prefix matches before checking. | 24 July 2018, 18:18:11 UTC |
| 29caae5 | brharrington | 21 July 2018, 00:01:13 UTC | additional logging for rejected datapoints (#857) Occasionally it is useful to sample this if the client is ignoring the response. | 21 July 2018, 00:01:13 UTC |
| 9597683 | brharrington | 20 July 2018, 19:18:30 UTC | remove deprecated usage of Class.newInstance (#856) Fixes #834. | 20 July 2018, 19:18:30 UTC |
| 2f3dc44 | brharrington | 20 July 2018, 18:27:56 UTC | add trace logging for input events (#855) This will make it easier to capture raw data coming from the server as well as the set of data sources that were being run on the evaluator. | 20 July 2018, 18:27:56 UTC |
| 3212798 | brharrington | 18 July 2018, 17:42:28 UTC | fix hash code for LongIntHashMap (#854) Before it was computing the murmur3 hash and then ignoring the result and using `Long.hashCode`. This change moves the computation of the murmur3 hash to a helper function and now LongIntHashMap uses that value. | 18 July 2018, 17:42:28 UTC |
| 18f8352 | brharrington | 18 July 2018, 14:27:38 UTC | update dependencies (#853) | 18 July 2018, 14:27:38 UTC |
| 96cd84f | Brian Harrington | 18 July 2018, 14:07:19 UTC | roaring bitmap 0.7.14 | 18 July 2018, 14:07:19 UTC |
| 30a9996 | Brian Harrington | 18 July 2018, 14:02:20 UTC | equalsverifier 2.4.8 | 18 July 2018, 14:02:20 UTC |
| cb0ca94 | Brian Harrington | 18 July 2018, 14:00:28 UTC | iep 1.2.5 | 18 July 2018, 14:00:28 UTC |
| 2ae8f66 | Brian Harrington | 18 July 2018, 13:51:02 UTC | aws-java-sdk 1.11.356 | 18 July 2018, 13:51:02 UTC |
| 449e05c | Brian Harrington | 18 July 2018, 13:44:45 UTC | joda-convert 2.1.1 | 18 July 2018, 13:44:45 UTC |
| 9a27ff9 | Brian Harrington | 18 July 2018, 13:43:42 UTC | spectator 0.72.0 | 18 July 2018, 13:43:42 UTC |
| da894a1 | Brian Harrington | 18 July 2018, 13:40:02 UTC | akka 2.5.14 Adds support for async DNS resolver. | 18 July 2018, 13:40:02 UTC |
| f76ba2d | brharrington | 11 July 2018, 15:49:28 UTC | handle monotonic values with multiple samples (#851) For cloudwatch if a monotonic value gets multiple samples within an interval it will cause a spike when looking at the sum. Use the max to see the largest sample within a given interval. Example metrics with this issue are the HSM network stats (InterfaceEth2OctetsInput). https://docs.aws.amazon.com/cloudhsm/latest/userguide/hsm-metrics-cw.html | 11 July 2018, 15:49:28 UTC |
| 49a05f2 | davecromberge | 11 July 2018, 02:33:16 UTC | record the total number of unique metrics as gauge (#850) The index stats now have provision for additional gauges that concern the index as a whole, not only key statistics. The first of such metrics is the total number of unique items, or metrics, that up to now has not been recorded. | 11 July 2018, 02:33:16 UTC |
| c393edb | brharrington | 10 July 2018, 05:21:50 UTC | config setting for window of time grouping (#849) Adds a config setting that can be used to set the number of buffers that will be maintained when grouping the datapoints into time buckets. The default is 1 which will flush the data as soon as data arrives for the next time frame. More buffers allows more time for data to accumulate, but means that it will take longer to emit data for a given time. | 10 July 2018, 05:21:50 UTC |
| f702809 | Matthew Johnson | 29 June 2018, 21:47:27 UTC | add cloudwatch rds FreeStorageSpace metric (#846) Fixes #845. | 29 June 2018, 21:47:27 UTC |
| 3287576 | brharrington | 22 June 2018, 23:19:15 UTC | fix name for AuroraBinlogReplicaLag (#844) AWS is inconsistent in the casing for BinLog between the disk usage and replica lag metrics. | 22 June 2018, 23:19:15 UTC |
| bac8e64 | brharrington | 22 June 2018, 22:57:57 UTC | add mapping for RDS replica lag (#843) The value is the number of seconds of lag. | 22 June 2018, 22:57:57 UTC |
| 5c3617a | brharrington | 21 June 2018, 20:53:39 UTC | add test case for `^*` behavior with matcher (#842) There was some inconsistency between the UI validation which uses the javascript regex library and the backend. Originally we suspected this was a bug and it was getting mapped to an IndexOf query, but that is not the case. The java regex engine has a slightly different interpretation. | 21 June 2018, 20:53:39 UTC |
| 95ffab7 | brharrington | 18 June 2018, 17:05:04 UTC | dependency updates (#840) | 18 June 2018, 17:05:04 UTC |
| 2adacdd | Brian Harrington | 18 June 2018, 16:48:17 UTC | spectator 0.70.2 Minor change needed for test case because spectator will no longer force gauge updates inline. For more information see Netflix/spectator#562. | 18 June 2018, 16:48:17 UTC |
| 5f96692 | Brian Harrington | 18 June 2018, 16:39:27 UTC | aws-java-sdk 1.11.339 | 18 June 2018, 16:39:27 UTC |
| 0b16cb7 | Brian Harrington | 18 June 2018, 16:14:07 UTC | akka-http 10.1.3 | 18 June 2018, 16:38:34 UTC |
| c9f8000 | Brian Harrington | 18 June 2018, 15:54:01 UTC | akka 2.5.13 | 18 June 2018, 15:54:01 UTC |
| 3fd4cb7 | Brian Harrington | 18 June 2018, 15:52:25 UTC | jackson 2.9.6 Has some additional gadget security fixes. | 18 June 2018, 15:52:25 UTC |
| a3a4ad1 | brharrington | 11 June 2018, 16:02:15 UTC | move metered queue config out of reference.conf (#839) Makes it easier for us to do comparison performance tests with other implementations. It can be included with the application config when needed. | 11 June 2018, 16:02:15 UTC |
| 5799522 | brharrington | 02 June 2018, 03:12:06 UTC | remove useless call (#836) The return value is never checked. | 02 June 2018, 03:12:06 UTC |
| 8de4f09 | brharrington | 02 June 2018, 03:10:48 UTC | do not show lines below lower bound with log scale (#835) Fixes #832. Before values below the lower bound would get mapped to a 0 and look like a line along the lower bound. If it is below the bound, then it should not show. | 02 June 2018, 03:10:48 UTC |
| 000eac0 | brharrington | 01 June 2018, 22:53:26 UTC | update travis to build on jdk10 (#833) Dropping jdk9 as there is no reason to use it over jdk10 at this point. One test case needed to be fixed because there is now a `java.version.date` system property and it was accessing `java.version` expecting a string instead of an object. The tests fail on OpenJDK because of font diffs. For more information see: http://mail.openjdk.java.net/pipermail/discuss/2017-November/004621.html Equals verifier is not yet compatible with jdk11 bytecode. Also some image diffs are present when using the early access OracleJDK 11. So for now The build will only run on OracleJDK 10. | 01 June 2018, 22:53:26 UTC |
| 2a75a48 | brharrington | 31 May 2018, 22:48:44 UTC | fix state isolation for streaming execution (#831) If an expression has multiple equivalent stateful sub-expressions, then the they would store the state to the same place since the key is the stateful expr instance. This changes the map used to look at the identity when performing the state lookup, so separate instances will have isolated state. We'll probably need to refactor this more thoroughly long term and either memoize equivalent sub-expressions during the evaluation or lookup based on position. | 31 May 2018, 22:48:44 UTC |
| 63e338b | brharrington | 30 May 2018, 20:01:23 UTC | honor legend flag for v2.json format image (#830) See #564 and #568 for more context. The included image would suppress the image. For some use-cases it is desirable to include it. The previous output can still be achieved by explicitly using `no_legend=1`. Main use-case right now is an explode view that shows the images, but has some dynamic filtering for outliers or missing data. This avoids the need to make multiple requests for the same data if the user requests to see the legends. | 30 May 2018, 20:01:23 UTC |
| 5984c7b | brharrington | 23 May 2018, 03:33:30 UTC | dependency updates (#829) | 23 May 2018, 03:33:30 UTC |
| 1a4fa06 | Brian Harrington | 23 May 2018, 03:21:57 UTC | sbt 1.1.5 | 23 May 2018, 03:21:57 UTC |
| a9b9f5d | Brian Harrington | 23 May 2018, 03:11:18 UTC | sbt-jmh 0.3.4 | 23 May 2018, 03:11:18 UTC |
| 85b98f4 | Brian Harrington | 22 May 2018, 23:46:56 UTC | akka 2.5.12 | 22 May 2018, 23:46:56 UTC |
| f7c436b | Brian Harrington | 22 May 2018, 23:46:28 UTC | RoaringBitmap 0.7.10 | 22 May 2018, 23:46:28 UTC |
| eb2863e | Brian Harrington | 22 May 2018, 23:40:50 UTC | equalsverifier 2.4.6 | 22 May 2018, 23:40:50 UTC |
| 75534c7 | Brian Harrington | 22 May 2018, 23:35:25 UTC | scala-logging 3.9.0 | 22 May 2018, 23:35:25 UTC |
| 3fc1601 | Brian Harrington | 22 May 2018, 23:12:08 UTC | spectator 0.68.0 and iep 1.2.3 | 22 May 2018, 23:26:39 UTC |
| 31b329d | Brian Harrington | 22 May 2018, 23:10:14 UTC | scalafmt 1.5.1 | 22 May 2018, 23:10:14 UTC |
| 16ed668 | brharrington | 16 May 2018, 00:02:31 UTC | set dstype for polled cloudwatch metrics (#827) It was not set before so all would get treated as rate. This led to some confusion for some of the S3 metrics where it appeared there was a massive drop in the amount of data stored in s3 after a brief gap. This drop was because the consolidation averaged with 0 for the NaN values since the values were considered rates. | 16 May 2018, 00:02:31 UTC |
| 45e6520 | brharrington | 30 April 2018, 19:50:12 UTC | update to scala 2.12.6 (#821) | 30 April 2018, 19:50:12 UTC |
| c624dbf | brharrington | 23 April 2018, 01:20:48 UTC | fix possible contract error for IndexStats (#820) If there are duplicates for the number of distinct values, then it is possible to get an error from the sort: ``` java.lang.IllegalArgumentException: Comparison method violates its general contract! at java.util.TimSort.mergeHi(TimSort.java:899) ~[?:1.8.0_162] at java.util.TimSort.mergeAt(TimSort.java:516) ~[?:1.8.0_162] at java.util.TimSort.mergeCollapse(TimSort.java:441) ~[?:1.8.0_162] at java.util.TimSort.sort(TimSort.java:245) ~[?:1.8.0_162] at java.util.Arrays.sort(Arrays.java:1438) ~[?:1.8.0_162] ``` This adds a test and fixes the comparison operation. | 23 April 2018, 01:20:48 UTC |
| 1a56948 | brharrington | 22 April 2018, 19:29:16 UTC | add helper for tracking index stats (#819) This provides a quick summary of some key stats for the index to see what is contributing the most to the number of distinct strings. In most cases we just use hive queries for digging into this, but these stats can be used before data is in hive and are always scoped to the specific index. In some cases where we already have rollup policies the hive data is raw from before the additional policies are enabled. | 22 April 2018, 19:29:16 UTC |
| 1621357 | Daniel Muino | 19 April 2018, 00:15:01 UTC | Add :pow operator (#818) Allows the user to use `math.pow` on time series expressions. `a,b,:pow` will return a new time series where the values are `a` taken to the power of `b`. | 19 April 2018, 00:15:01 UTC |
| 995c807 | brharrington | 10 April 2018, 17:20:42 UTC | switch to `skip in publish := true` (#817) The old no packaging hack is no longer needed on recent versions of sbt and sbt-bintray. | 10 April 2018, 17:20:42 UTC |
| 6bec417 | brharrington | 06 April 2018, 23:10:50 UTC | iep 1.2.2 (#815) This is mostly just to trigger a change to verify the release. The previous attempt at 1.6.0-rc.10 failed because it was still trying to publish 1.6.0-rc.9. The theory is that this is because they were both tagging the same commit. | 06 April 2018, 23:10:50 UTC |
| 44e2f58 | brharrington | 04 April 2018, 16:13:40 UTC | ensure batches are regularly pushed (#814) For smaller deployments it would take time to fill up the entire batch causing a significant delay in the publishing. Now it will flush based on the age even if the batch is not yet full. | 04 April 2018, 16:13:40 UTC |
| 6cb745b | Daniel Muino | 03 April 2018, 21:46:18 UTC | Collect maximumUsedTranactionIds from cloudwatch (#813) This will help detect Transaction ID Wraparound in postgresql databases. See: https://aws.amazon.com/blogs/database/implement-an-early-warning-system-for-transaction-id-wraparound-in-amazon-rds-for-postgresql/ for details | 03 April 2018, 21:46:18 UTC |
| 871daf1 | brharrington | 03 April 2018, 17:30:12 UTC | reduce number of nodes in demo dataset (#812) This is used for test cases and docs. Reducing the number of nodes significantly improves the time for running tests or generating docs that use longer time ranges. | 03 April 2018, 17:30:12 UTC |
| 101260f | brharrington | 02 April 2018, 22:54:59 UTC | update dependencies (#811) | 02 April 2018, 22:54:59 UTC |
| a1e84d6 | Brian Harrington | 02 April 2018, 22:27:23 UTC | sbt-updates 0.3.4 | 02 April 2018, 22:27:23 UTC |
| 53d309a | Brian Harrington | 02 April 2018, 22:16:14 UTC | sbt-release 1.0.8 | 02 April 2018, 22:16:14 UTC |
| 0669a51 | Brian Harrington | 02 April 2018, 22:03:47 UTC | sbt-bintray 0.5.4 | 02 April 2018, 22:03:47 UTC |
| 85b77f1 | Brian Harrington | 02 April 2018, 21:59:05 UTC | sbt 1.1.2 | 02 April 2018, 21:59:05 UTC |
| ce2d7ea | Brian Harrington | 02 April 2018, 21:49:48 UTC | jackson 2.9.5 | 02 April 2018, 21:49:48 UTC |
| 05b9233 | Brian Harrington | 02 April 2018, 21:37:29 UTC | joda-convert 2.0.1 | 02 April 2018, 21:37:29 UTC |
| d1009fd | Brian Harrington | 02 April 2018, 21:34:36 UTC | roaringbitmap 0.7.9 | 02 April 2018, 21:34:36 UTC |
| 2c91df9 | Brian Harrington | 02 April 2018, 21:27:00 UTC | caffeine 2.6.2 | 02 April 2018, 21:27:00 UTC |
| 7d33e58 | Brian Harrington | 02 April 2018, 21:25:46 UTC | typesafe config 1.3.3 | 02 April 2018, 21:25:46 UTC |
| a0b9a59 | Brian Harrington | 02 April 2018, 21:19:43 UTC | scala-logging 3.8.0 | 02 April 2018, 21:19:43 UTC |
| 360486e | Brian Harrington | 02 April 2018, 18:25:40 UTC | akka-http 10.1.1 Adds mavenCentral to the list of resolvers because the `pom.sha1` file for akka-http_2.11 10.1.1 is broken in jcenter. For more details see akka/akka-http#1972. | 02 April 2018, 21:12:29 UTC |
| ad68bcd | Brian Harrington | 02 April 2018, 18:23:02 UTC | akka 2.5.11 | 02 April 2018, 18:23:02 UTC |
| 35eadde | Brian Harrington | 02 April 2018, 18:08:03 UTC | aws-java-sdk 1.11.299 | 02 April 2018, 18:08:03 UTC |
| 65a38bb | Brian Harrington | 02 April 2018, 18:03:21 UTC | log4j 2.11.0 | 02 April 2018, 18:03:21 UTC |
| 434324b | Brian Harrington | 02 April 2018, 18:00:50 UTC | iep 1.2.1 | 02 April 2018, 18:00:50 UTC |
| f1717c8 | Brian Harrington | 02 April 2018, 17:55:54 UTC | spectator 0.63.0 | 02 April 2018, 17:55:54 UTC |
| 2a328f2 | brharrington | 23 March 2018, 21:34:57 UTC | preserve offset information for :percentiles (#810) Before it was getting lost if the percentile expression was written back out to a string. Fixes #809. | 23 March 2018, 21:34:57 UTC |
| 37d24f3 | brharrington | 23 March 2018, 05:07:41 UTC | support rendering pre-evaluated data (#808) Updates the Grapher lib so it can be used to render data that has already been evaluated. The previous methods that accept the input data or data expr map have been renamed to evalAndRender. | 23 March 2018, 05:07:41 UTC |
| 464f206 | brharrington | 19 March 2018, 16:28:55 UTC | indicate DST transition on time axis (#807) A frequent question after a DST transition is why the time axis labels are off. This change indicates the DST transition by changing the first major tick label after to show `+1H` or `-1H`. The color is red to make it standout from other tick lables on the same axis. | 19 March 2018, 16:28:55 UTC |
| 2c2f0ac | brharrington | 16 March 2018, 15:59:37 UTC | avoid config loading for demo dataset (#806) This is used for generating fake data for the purposes of testing or creating documentation examples. The config load is causing problems due to classloader issues when running as part of build plugin. Not sure why, but for this use-case it can be avoided. | 16 March 2018, 15:59:37 UTC |
| 2d6dd9e | brharrington | 16 March 2018, 01:58:19 UTC | docs: expand description of expr scoped palettes (#805) | 16 March 2018, 01:58:19 UTC |
| cc31e29 | brharrington | 16 March 2018, 00:03:29 UTC | docs: describe common use-case for clamp-max (#804) | 16 March 2018, 00:03:29 UTC |
| a273609 | brharrington | 09 March 2018, 21:30:25 UTC | akka-http 10.1.0 (#803) Changes default connection pool implementation. So far the new implementation looks find based on test deployment with the RC2 release. | 09 March 2018, 21:30:25 UTC |
| bf38a75 | brharrington | 07 March 2018, 23:48:36 UTC | eval: reject datapoints that are from the future (#802) If we get timestamps from the the future, then it can mess up the time based grouping for the input. Before the buckets for the grouping would adjust to the future time and current data would get dropped as being too old. | 07 March 2018, 23:48:36 UTC |
| 218cbd8 | brharrington | 07 March 2018, 05:57:04 UTC | add support for palette scoped to an expression (#801) This change will allow a palette to be used for a particular expression. This can be used with group by operations to use a given color scheme per group, for example, success codes are green and errors are reds. Fixes #800. | 07 March 2018, 05:57:04 UTC |
| 7cfde45 | brharrington | 28 February 2018, 20:55:02 UTC | remove bad api from static config (#799) The RenderApi never really worked and was removed as part of #797. | 28 February 2018, 20:55:02 UTC |
| b3180d9 | brharrington | 27 February 2018, 00:48:30 UTC | initial work to split rendering from webapi (#797) There are a number of use-cases where it is convenient to be able to render a chart based on the uri without going through the webapi. Examples: - Alerting through LWC with a certain buffer for the data being streamed through. If an alert triggers, then it is desirable to be able to include context in the email or other notification. The graph image helps the user understand why it fired. - Spark queries over the data in S3. We want to be able to let a user use the exact same graph uri they use for the main backend and run it on spark. This is an initial step and we'll try it out a bit to see what changes make sense. A subset of the tests from the GraphApiSuite have been moved over. Most of the rest will follow later on so we can ensure that there isn't anything being added in the graph layer. | 27 February 2018, 00:48:30 UTC |
| a9f3bf0 | brharrington | 26 February 2018, 23:37:10 UTC | add mapping for dynamo-db throttle events (#796) These appear to only be available at the table level. The error for request throttled would be needed to see how it impacts particular operations. | 26 February 2018, 23:37:10 UTC |
| 4f1ea8a | brharrington | 26 February 2018, 01:20:21 UTC | fix removal of key that is not in SmallHashMap (#795) Before it would fail with: ``` requirement failed: data array is full ``` | 26 February 2018, 01:20:21 UTC |
| f82fff4 | brharrington | 24 February 2018, 18:48:42 UTC | docs: fix end tag for example tables (#794) Change `<s/tr>` to `</tr>`. | 24 February 2018, 18:48:42 UTC |
| 849ef81 | brharrington | 23 February 2018, 23:51:48 UTC | simplify string mapping for NamedRewrite (#793) If the final grouping for display varies from the final grouping of the eval, then the rewrite is applying a secondary grouping. Otherwise just apply the rewrite operator. Fixes #791. | 23 February 2018, 23:51:48 UTC |
| e13de86 | brharrington | 23 February 2018, 23:16:31 UTC | fix additional case of losing offset (#792) The group by operation could lose the offset if it was not encoded as part of the display expression. | 23 February 2018, 23:16:31 UTC |
| 070571f | brharrington | 23 February 2018, 16:57:01 UTC | fix offset for aggregate function rewrites (#790) Before if the display expression for a rewrite was a Query type it would not correctly get the offset applied. Fixes #788. | 23 February 2018, 16:57:01 UTC |
| 6f5567e | brharrington | 15 February 2018, 03:12:23 UTC | re-enable scala 2.11 build (#787) It was dropped in #600. There are some use-cases where it would be helpful to use the latest version from Spark which does not yet support 2.12. We'll revisit for Atlas 1.7. | 15 February 2018, 03:12:23 UTC |
| 9680cdd | brharrington | 15 February 2018, 00:44:06 UTC | add format target (#785) Adds a format target to the makefile to make it easier to run both the license formatting and scala format. | 15 February 2018, 00:44:06 UTC |
| 87ae556 | brharrington | 15 February 2018, 00:43:46 UTC | update dependencies (#786) | 15 February 2018, 00:43:46 UTC |
| f0f0ec6 | Brian Harrington | 15 February 2018, 00:26:07 UTC | sbt-jmh 0.3.3 | 15 February 2018, 00:26:07 UTC |
| 8e0033c | Brian Harrington | 15 February 2018, 00:24:17 UTC | aws-java-sdk 1.11.277 | 15 February 2018, 00:24:17 UTC |
| 976041f | Brian Harrington | 15 February 2018, 00:17:31 UTC | equalsverifier 2.4.3 | 15 February 2018, 00:17:31 UTC |
| 9e8dbc6 | Brian Harrington | 15 February 2018, 00:14:15 UTC | caffeine 2.6.1 | 15 February 2018, 00:14:15 UTC |
| a4121e1 | Brian Harrington | 14 February 2018, 23:44:33 UTC | RoaringBitmap 0.7.0 | 14 February 2018, 23:58:21 UTC |
| e234012 | Brian Harrington | 14 February 2018, 23:38:26 UTC | scalatest 3.0.5 | 14 February 2018, 23:38:26 UTC |
| 0bd0580 | Brian Harrington | 14 February 2018, 23:35:24 UTC | spectator 0.61.0 | 14 February 2018, 23:35:24 UTC |
| 821d381 | Brian Harrington | 14 February 2018, 23:34:08 UTC | iep 1.2.0 | 14 February 2018, 23:34:08 UTC |
| f3f9f70 | Brian Harrington | 14 February 2018, 23:32:12 UTC | sbt 1.1.1 | 14 February 2018, 23:32:12 UTC |
| 6e80e2d | brharrington | 14 February 2018, 17:57:37 UTC | fix use of multi-level group by with rewrites (#784) Fixes #584. Before this would result in errors if the rewrite resulted in a non-grouped time series expression. | 14 February 2018, 17:57:37 UTC |
| 24eedd2 | brharrington | 13 February 2018, 20:41:58 UTC | add group by keys to time series message (#783) Fixes #754. Exposes the set of keys that were used explicitly as part of a group by in the time series messages emitted by fetch and streaming evaluation. This is useful for things like canary analysis that generate a set of lines for multiple clusters and need a reliable way to find the correlated entries from the baseline and canary. Since the tags will contain exact matches from the query as well as keys used in the group by, this change will allow the user to extract just the keys from the group by and ignore other exact match tags coming from the scope used on the query. | 13 February 2018, 20:41:58 UTC |
