| 4634131 | Dan LaRocque | 26 November 2014, 14:35:07 UTC | Updating some Gremlin snippets in the docs for 0.5 The most important change in this commit is replacement of some Gremlin `(iterable of vertices).type` calls with `(iterable of vertices)*.getVertexLabel()`. I also ran through the getting-started steps in the intro again on 0.5 and copy-pasted the output as it appears today: type=foo no longer appears in `map` output, calling `map` on a single vertex lists each property on a separate line (used to be all on one line enclosed by curly braces apparently), and edge toStrings are formatted slightly differently. The rest of the changes should just account for different vertex IDs than used to be assigned in the getting-started scenario. (cherry picked from commit f39fae8fdc1fdd1ab1e2531efb0ccfa4a8179c48) | 26 November 2014, 14:43:25 UTC |
| 99949bd | Brad Anderson | 11 November 2014, 18:06:58 UTC | docs patch (cherry picked from commit c17123dbd36c92ecf3d5b0ad79d2a84bff9fd36b) | 25 November 2014, 20:22:17 UTC |
| 2d241c5 | Dan LaRocque | 25 November 2014, 18:35:42 UTC | Tweak traversal as directed by @dkuppitz New traversal snippet uses the start-dot operator: http://groovy.codehaus.org/Collections#Collections-Gettingefficientwiththestar-dot%27*.%27operator (cherry picked from commit b804fd78ac256d7b95fa90201a7935bdaf94d95f) | 25 November 2014, 18:37:41 UTC |
| 9b6056d | Dan LaRocque | 24 November 2014, 18:18:59 UTC | [maven-release-plugin] prepare release 0.5.2 | 24 November 2014, 18:18:59 UTC |
| e721f73 | Dan LaRocque | 24 November 2014, 18:17:59 UTC | Updating titan.compatible.versions prop | 24 November 2014, 18:17:59 UTC |
| 322b918 | Dan LaRocque | 24 November 2014, 18:14:04 UTC | Add sleep to testIndexQueryWithScore ES rebuilds indexes asynchronously, but testIndexQueryWithScore commits and then immediately tries to search on what was just committed. Maybe we should change the ES test config to force an inefficient full refresh on every write for testing purposes. For now, I'm adding a sleep to the test instead. | 24 November 2014, 18:17:19 UTC |
| ea126c7 | Dan LaRocque | 24 November 2014, 16:49:10 UTC | Merge branch 'issue_808' into titan05 | 24 November 2014, 16:49:10 UTC |
| 2182eb8 | Dan LaRocque | 24 November 2014, 14:10:51 UTC | Set 0.5.2 release date in upgrade instructions | 24 November 2014, 14:10:51 UTC |
| 1e11c64 | Dan LaRocque | 24 November 2014, 14:09:54 UTC | Make release.sh push HTML documentation to S3 | 24 November 2014, 14:09:54 UTC |
| 9fc802d | Dan LaRocque | 24 November 2014, 14:08:52 UTC | Bring release instructions up-to-date | 24 November 2014, 14:09:27 UTC |
| 16f1fae | Dan LaRocque | 24 November 2014, 13:54:47 UTC | Tweak gh-pages-update.sh This script generates a new gh-pages index.html and copies javadocs for the release to /javadoc/$VERSION/ and /javadoc/current/. This commit makes its detection of whether the gh-pages branch already exists more robust by using a git plumbing command instead of checking file existence in the .git directory. Also fix outdated wording talking about wikidoc in one of the messages it prints to console. | 24 November 2014, 13:54:47 UTC |
| 9dcf945 | Dan LaRocque | 24 November 2014, 13:46:38 UTC | READEish file updates for 0.5.2 | 24 November 2014, 13:46:38 UTC |
| 6205f7f | Dan LaRocque | 24 November 2014, 13:23:03 UTC | Remove Maps.immutableEnumMap call For #844. Guava 11 doesn't have this method, so it throws NoSuchMethodError on Hadoop clusters. Changed to ImmutableMaps.copyOf. | 24 November 2014, 13:23:03 UTC |
| 33fe2f0 | Pavel Yaskevich | 24 November 2014, 08:48:41 UTC | use "q=" to set conditions instead of "fq=" for Solr (issue #808) | 24 November 2014, 08:48:41 UTC |
| 34002ea | Dan LaRocque | 23 November 2014, 13:43:39 UTC | Support titan.hadoop.output.location Fix #822 | 23 November 2014, 13:45:08 UTC |
| 66a60a9 | Dan LaRocque | 23 November 2014, 13:42:36 UTC | Adding test of titan.hadoop.output.location For #822. output.location was recently removed and jobdir.location added instead, so this test of output.location currently fails. Reimplementation of the expected (pre-0.5) output.location behavior coming next. | 23 November 2014, 13:44:18 UTC |
| 274af8b | Dan LaRocque | 23 November 2014, 13:13:56 UTC | Merge branch 'titan05' of github.com:thinkaurelius/titan into titan05 | 23 November 2014, 13:13:56 UTC |
| 5a853d7 | Matthias Broecheler | 23 November 2014, 03:03:29 UTC | Support for Contain.IN in queries-centric queries with test coverage. Addresses the second part of #679. Fixes #679. | 23 November 2014, 03:03:29 UTC |
| 8fe79b1 | Matthias Broecheler | 23 November 2014, 00:29:57 UTC | Support for Contain.IN in Vertex-centric queries with test coverage. Addresses the first part of #679 | 23 November 2014, 00:29:57 UTC |
| 8d543a9 | Dan LaRocque | 22 November 2014, 22:27:51 UTC | Fix test excludes in titan-hadoop-parent This change should have been part of 856a84ee55aa106d29db3d6ade540b7ca627dd57 | 22 November 2014, 22:27:51 UTC |
| fb704eb | Matthias Broecheler | 22 November 2014, 04:50:52 UTC | Merge branch 'titan05' of https://github.com/thinkaurelius/titan into titan05 | 22 November 2014, 04:50:52 UTC |
| cfbe4c8 | Matthias Broecheler | 22 November 2014, 04:50:40 UTC | Added a test case to verify that concurrently updating low-selectivity index entries while reading from them causes result duplication because of query limit adjusting. Added configuration option to disable this behavior. Fixes #839 | 22 November 2014, 04:50:40 UTC |
| 90c2cd4 | Dan LaRocque | 21 November 2014, 20:29:38 UTC | Merge branch 'titan05' of github.com:thinkaurelius/titan into titan05 | 21 November 2014, 20:29:38 UTC |
| f8f8204 | Matthias Broecheler | 21 November 2014, 19:39:28 UTC | Fixes #782 and adds some test coverage for VertexList implementations. | 21 November 2014, 19:39:28 UTC |
| 856a84e | Dan LaRocque | 21 November 2014, 19:34:36 UTC | Add test that reads a list-valued prop with Faunus This test is modeled on #807. #807 is reproducible on 0.5.1 but has been fixed in titan05 (0.5.2-SNAPSHOT) at the time of this commit. This was probably fixed as part of earlier Faunus NPE issue fixing, but I haven't bisected and don't know exactly which change did it. This commit adds a test which would catch a regression on #807. | 21 November 2014, 19:34:36 UTC |
| 5e25600 | Matthias Broecheler | 21 November 2014, 19:01:25 UTC | Merge branch 'titan05' of https://github.com/thinkaurelius/titan into titan05 | 21 November 2014, 19:01:25 UTC |
| b9aca32 | Dan LaRocque | 21 November 2014, 18:57:41 UTC | Support Astyanax's setLocalDatacenter option Fixes #841 | 21 November 2014, 18:57:41 UTC |
| 1c37382 | Dan LaRocque | 21 November 2014, 18:55:44 UTC | Make STG's shutdown hook avoid removing itself This is for #842. The shutdown hook now carries out all shutdown steps except hook removal. The shutdown method on STG does everything the hook does plus hook removal. | 21 November 2014, 18:55:44 UTC |
| 018827a | Dan LaRocque | 21 November 2014, 18:45:34 UTC | Fix `gremlin.sh -l <LOGLEVEL>` I inadvertently broke this with 1b23ce606090eb7e9407e63e26bf8e4e2418ac42. Prior to that commit, gremlin.sh passed the "-l FOO" arguments into the Console main class, but the Console main class ignored them. That commit made the Console main class stop ignoring them and try to interpret "-l" as the path to an init file. This commit tweaks gremlin.sh to shift those "-l FOO" arguments off the bottom of the positional parameter set after processing, so that they aren't passed to the Console main class. | 21 November 2014, 18:48:11 UTC |
| d282b99 | Dan LaRocque | 21 November 2014, 17:02:59 UTC | Removing groovy-all from titan-test No longer necessary due to 18f44383dc18968930bac6c36822e0b68012d8c6 | 21 November 2014, 17:02:59 UTC |
| 18f4438 | Dan LaRocque | 21 November 2014, 15:16:51 UTC | Move shaded Groovy + ASM classes to external jar This removes the shaded copy of Groovy 1.8.9 and ASM 3.2 introduced to titan-core in bfcc99dbd71e90f4a2a17b24cf02ab484bf89396. That worked fine in the Maven build, tests, and in Rexster and gremlin.sh testing from the zip archives (in other words, substantially everywhere), but IDEA refused to compile the Groovy classes in this configuration. The Maven build worked; only IDEA failed to build. It seemed to ignore the shade plugin config in titan-core, manually compiling just titan-core/src/main but leaving out the Groovy and ASM classes, resulting in NoClassDefFound on groovy.lang.GroovyObject when building the other modules. It's possible to work around this IDEA limitation by manually adding Groovy 1.8.9 as a global library or as a project library in all the affected modules. IDEA does build the project in that case. But I don't like that approach, since it means IDEA and Maven tests will use different copies of the Groovy runtime. In an attempt to placate IDEA and while trying to keep testing as uniform as possible between Maven and IDEA, I pushed the shaded Groovy + ASM classes out to their own artifact now published to central, named groovy-shaded-asm. This moots IDEA's inability to comprehend the shade plugin config in titan-core, since titan-core is no longer responsible for shading (it's done in this new artifact). This is all related to #784. | 21 November 2014, 15:16:51 UTC |
| de721be | Dan LaRocque | 21 November 2014, 09:12:13 UTC | Add groovy-all titan-test dependency for IDEA This new dependency has scope=provided. Prior to this commit, Maven built the project without error, but IDEA seemed to get confused by some of the changes in #784, and refused to build titan-test with the error message "no groovy library is defined for module". | 21 November 2014, 09:12:13 UTC |
| e9cfca6 | Matthias Broecheler | 21 November 2014, 00:21:51 UTC | Merge branch 'titan05' of https://github.com/thinkaurelius/titan into titan05 | 21 November 2014, 00:21:51 UTC |
| dfb5bea | Matthias Broecheler | 21 November 2014, 00:21:43 UTC | Intermediate commit of test for VertexLists | 21 November 2014, 00:21:43 UTC |
| c3cc603 | Matthias Broecheler | 21 November 2014, 00:20:55 UTC | Fixed java verison in pom | 21 November 2014, 00:20:55 UTC |
| d80c9d8 | Matthias Broecheler | 20 November 2014, 19:56:26 UTC | Fixes #783 | 20 November 2014, 19:56:26 UTC |
| bfcc99d | Dan LaRocque | 20 November 2014, 09:02:44 UTC | Use Groovy 1.8.9 with shaded ASM 3.x This is part of #784. This effectively reverts b1635dcaf0754dcbf1d8dac84e4795b5c6f235ea, which upgraded from Groovy 1.8.9 to 2.3.7. This commit also configures the shade plugin in titan-core to hide Groovy 1.8.9's required copy of ASM 3.x under the package prefix com.thinkaurelius.shaded.groovy.asm. Since the Groovy 1.8.9 classfiles also need to be rewritten to use the shaded package name at every ASM type reference and method callsite, titan-core now includes a copy of Groovy 1.8.9. The package for Groovy's classes is unchanged to maintain compatibility with client code in Rexster and Gremlin-Groovy. Only ASM has moved. Furthermore, this commit adds a new maven-enforcer-plugin execution in the top-level pom. This execution bans dependencies on the ASM 3.x artifacts. ASM 4.x+ is still allowed (since it uses a different groupId than the 3.x series). If this enforcer execution finds an ASM 3.x dep, it will fail the Maven build. | 20 November 2014, 09:02:44 UTC |
| db612b5 | Dan LaRocque | 19 November 2014, 16:07:54 UTC | Remove track-state/paths=true from Hadoop cfgs Grepping recursively for "pipeline.track" in titan-dist/src where the master copies of the example configs live now produces no hits. The defaults for the track-state and track-paths options as defined in TitanHadoopConfiguration also remain false. Fixes #823 | 19 November 2014, 16:07:54 UTC |
| f01d045 | Dan LaRocque | 19 November 2014, 13:55:21 UTC | Making INDEX_DIRECTORY MASKABLE INDEX_DIRECTORY was GLOBAL_OFFLINE prior to this commit. I think MASKABLE will be a relatively safe change for 0.5.2 (LOCAL would not be safe). This should be allowed to vary to support nodes with heterogeneous disk layouts. | 19 November 2014, 13:57:40 UTC |
| 4ba6d1f | Dan LaRocque | 19 November 2014, 13:52:05 UTC | Adding config option for custom MR job classpaths This is related to #800. This commit adds the option "titan.hadoop.mr.classpath-configurer". When present, it's interpreted as the full package.classname of a class implementing JobClasspathConfigurer which has a nullary constructor. This takes precedence over the mapredjar and distcache configurer mechanisms already present in Faunus. | 19 November 2014, 13:52:05 UTC |
| a6e6d75 | Dan LaRocque | 19 November 2014, 13:48:12 UTC | Delete Cassandra data between Expect ITs 8fa00662730ec74089dc4dd604cb247f4dadafa9 introduced a new test, forming a pair of Cassandra tests that could interfere with each other by using the same shared embedded Cassandra instance without dropping data between methods. This adds a @Before that drops those data. | 19 November 2014, 13:48:12 UTC |
| 063d3a3 | Dan LaRocque | 19 November 2014, 13:46:02 UTC | Increasing timeout in Faunus-GraphSON test I've seen this spuriously fail when the machine was under load during the tests. Increasing the timeout. | 19 November 2014, 13:46:02 UTC |
| 8b0a078 | Dan LaRocque | 19 November 2014, 13:44:12 UTC | Fix bug in shorthand argument interpretation a2149cec1d2d9b52b1bad596ef8cb1accdd2f3db broke a static initializer cycle triggered by TitanFactory.open('backend:arg') shorthand invocation, but it also added a bug that broke interpretation the 'arg' part of the shorthand. This is related to #831. | 19 November 2014, 13:44:12 UTC |
| b1635dc | Dan LaRocque | 19 November 2014, 13:25:04 UTC | Attempt upgrade to Groovy 2.3.7 This is part of #784. This commit overrides the Groovy version from 1.8.9 (what TP 2.5.0 compiled against and wants at runtime, but which depends on unshaded ASM 3.x) to 2.3.7, which uses a shaded copy of ASM built with the jarjar utility. This is not ideal. I recompiled gremlin-groovy against 2.3.7 and ran the test suite, which passed, but the Console class required some changes on account of the jline upgrade (1.x to 2.x involved some backwards-incompatible changes to jline's history package). There could be complications here that I'm not seeing yet. | 19 November 2014, 13:43:04 UTC |
| f47ee13 | Dan LaRocque | 19 November 2014, 13:08:37 UTC | Upgrading Avro version This is part of #784. The Avro version managed in the top-level pom prior to this commit, 1.4.1, depended on paranamer, which in turn depended on an unshaded copy of ASM 3.2: [INFO] +- org.apache.hadoop:hadoop-client:jar:2.2.0:compile [INFO] | \- org.apache.hadoop:hadoop-common:jar:2.2.0:compile [INFO] | \- org.apache.avro:avro:jar:1.4.1:compile (version managed from 1.7.4) [INFO] | \- com.thoughtworks.paranamer:paranamer-ant:jar:2.2:compile [INFO] | \- com.thoughtworks.paranamer:paranamer-generator:jar:2.2:compile [INFO] | \- (asm:asm:jar:3.1:compile - version managed from 3.2; omitted for duplicate) This commit updates to Avro 1.7.4, which depends on a different paranamer version that apparently no longer requires unshaded ASM, though I don't know whether that's because paranamer stopped using ASM or whether they've got a shaded and repackaged copy of it somewhere in their artifacts. | 19 November 2014, 13:08:37 UTC |
| e9f5edd | Dan LaRocque | 19 November 2014, 13:04:29 UTC | Update Jersey version from 1.17 to 1.18.2 This is related to #784. 1.17 depends on an unshaded copy of ASM 3.2. 1.18.2 drops the unshaded ASM dependency in favor of carrying its own shaded copy of the ASM classfiles. I'm not certain Rexster will tolerate this. | 19 November 2014, 13:04:29 UTC |
| b48f1e0 | Dan LaRocque | 19 November 2014, 12:45:37 UTC | Cleanup HBase version management The purpose of this commit is twofold: * Centralize HBase version management. The target 0.98.x version was duplicated in several places prior to this commit. Now the top-level pom property ${titan098.core.version} specifies the HBase 0.98 version (without the -hadoop2 suffix), which is appended in ${titan098.version}. * Update to the latest HBase bugfix releases. 0.98 moved from .2 and .3 to .8, and 0.94 moved from .21 to .25. 0.96 does not appear to have new upstream bugfixes and stays at .2. | 19 November 2014, 13:02:08 UTC |
| b4eee85 | Dan LaRocque | 18 November 2014, 18:20:07 UTC | Move Faunus Graph instantiation into RR.initialize I kind of implemented this in 5dd656e050cf1f31580a4425e51b3da272d9bfcb, but I cheated by opening the graph in InputFormat.createRecordReader rather than RecordReader.initialize. I'm not sure whether the contract of createRecordReader makes that safe. This commit moves graph instantiation into RecordReader.initialize in an attempt to remove the uncertainty. Related to #817 | 18 November 2014, 18:22:38 UTC |
| c12d1eb | Dan LaRocque | 18 November 2014, 17:50:45 UTC | Change some github.com urls from master to titan05 This is prompted by #803. I did a grep through all of the docs (recursive) for blob/master and found two hits. A more elegant solution would probably involve an AsciiDoc property that's centrally controlled to avoid repetition, but with only two references, I don't want to go overboard. Changed both to titan05 equivalents and checked that the links were still live. | 18 November 2014, 17:50:45 UTC |
| e8e7201 | Dan LaRocque | 18 November 2014, 17:35:25 UTC | Upgrade notice about Cassandra dirs A warning to users of titan.sh in 0.5.0 and 0.5.1 porting data forward to 0.5.2 that the script's working directory no longer affects the location of Cassandra data directories Related to #825 | 18 November 2014, 17:36:52 UTC |
| b2ad7e8 | Dan LaRocque | 18 November 2014, 17:14:43 UTC | Make the Cassandra working dir fixed The cassandra.yaml file used by titan.sh has long used relative paths for data, commitlogs, and saved_caches, but since moving away from the embeddedcassandra backend in favor of a standalone cassandra daemon, we can no longer rely on STORAGE_DIRECTORY being interpreted relative to the Titan configuration file. This commit changes cassandra.in.sh to cd to $CASSANDRA_HOME (typically the directory containing titan's bin/ directory, or the grandparent of titan.sh) before starting Cassandra. Fixes #825 | 18 November 2014, 17:14:43 UTC |
| a0a726d | Dan LaRocque | 18 November 2014, 16:55:06 UTC | Port old Hadoop-S3 Titan wikidoc page to AsciiDoc This commit introduces an AsciiDoc-ified copy of the content on https://github.com/thinkaurelius/faunus/wiki/Using-Amazon-S3-for-Storage as a new chapter (42) under the Hadoop part (IV) of the manual. | 18 November 2014, 16:55:06 UTC |
| db0984d | Dan LaRocque | 18 November 2014, 16:42:05 UTC | Fix misplaced semicolons in docs on TitanFactory Besides fixing syntax errors in the BDB page's example TitanFactory invocation, I appended a semicolon after the `open()` call in the HBase and Cassandra example TitanFactory invocations to make them copy-paste friendlier. All of these example code snippets declare Java as their markup language in the docs source and use Java-style explicit type declaration before the variable name, so I think this is appropriate. I didn't touch the inmemory backend's TitanFactory snippet since the docs explicitly describe the snippet as a Gremlin fragment. Fixes #801 | 18 November 2014, 16:44:07 UTC |
| 25cf8f5 | Dan LaRocque | 18 November 2014, 16:39:48 UTC | Fix "todo" download link in documentation Fixes #803 | 18 November 2014, 16:41:34 UTC |
| a2149ce | Dan LaRocque | 18 November 2014, 16:03:40 UTC | Breaking static initializer cycles GraphDatabaseConfiguration no longer contains references to Backend. Those references have been replaced by two new self-contained enums representing storage managers and index backends. Fixes #831. | 18 November 2014, 16:03:40 UTC |
| c546e13 | Dan LaRocque | 18 November 2014, 14:30:16 UTC | Failing test executing a static initializer cycle This test triggers a unsafe cycle of dependencies between static initializers in two classes, Backend and GraphDatabaseConfiguration, by invoking TitanFactory.open with a shorthand parameter first thing after the JVM starts up. This cycle is not triggered if a file is passed in. This test is for #831. | 18 November 2014, 16:00:50 UTC |
| 5dd656e | Dan LaRocque | 18 November 2014, 13:43:47 UTC | Moving TitanGraph instantiation into RecordReaders This should address #817. According to that issue, Spark (and maybe MR?) creates an InputFormat when starting a workload just to get the splits, then throws the IF away. Creating a graph in IF.setConf therefore lets an unclosed graph go out-of-scope in this scenario. This commit implements @skavulya's suggestion to move graph instantiation into RecordReader.initialize for both the HBase and Cassandra RRs implementations. | 18 November 2014, 13:47:46 UTC |
| 1b23ce6 | Dan LaRocque | 18 November 2014, 10:10:13 UTC | Make gremlin.sh process an arg as a gremlin script Fixes #824 | 18 November 2014, 10:10:13 UTC |
| 60672b0 | Dan LaRocque | 18 November 2014, 09:42:55 UTC | Add test of gremlin.sh gremlin-script.groovy This test checks that gremlin.sh interprets a single command-line argument, when present, as the path to a script file to be evaluated and whose additions to the scriptengine variable binding will be available at the prompt eventually presented to the user. The sample script used in the test sets and checks these four variables: test_i = 2 + 2 test_g = new TinkerGraph() test_hadoop_class = HadoopFactory.class test_titan_class = TitanFactory.class This is for #824 | 18 November 2014, 09:53:23 UTC |
| 6117813 | Dan LaRocque | 18 November 2014, 09:38:10 UTC | Move titan-dist test resources to test-classes The titan-dist pom prior to this commit copied test resources to the target/classes directory. This doesn't get included in any of Titan's release assemblies right now, but it's fragile. Moved these resources -- none of which should ever be included in a release archive -- to target/test-classes to better reflect their function and reduce the risk of accidental inclusion in future releases. | 18 November 2014, 09:38:10 UTC |
| 8fa0066 | Dan LaRocque | 18 November 2014, 06:37:37 UTC | Adding test for #786 This test runs in the aurelius-release profile. I implemented it as an Expect-based integration test in titan-dist because that was the most straightward approach within the existing test harness that allowed for restarting the JVM in the middle of the test logic, which is necessary to catch a regression on this particular bug. | 18 November 2014, 06:37:37 UTC |
| 502528c | Dan LaRocque | 18 November 2014, 04:35:02 UTC | Update old RDF config value in example Hadoop props The config value "n-triples" in rdf-input.properties should have been N_TRIPLES. | 18 November 2014, 04:35:02 UTC |
| 62d4591 | Dan LaRocque | 18 November 2014, 04:33:36 UTC | Put context.getConfiguration calls behind compat Converting calls to getConfiguration() on Hadoop Context instances to DEFAULT_COMPAT.getContextConfiguration(<instance>). Fixes #835. | 18 November 2014, 04:33:36 UTC |
| 949b837 | Dan LaRocque | 18 November 2014, 04:30:56 UTC | Adding test for #835 This fails under Hadoop 1 (but not 2) due to the bug reported in #835. Fix up next. | 18 November 2014, 04:33:10 UTC |
| aa1057d | Dan LaRocque | 15 November 2014, 07:14:32 UTC | Improve Hadoop job output dir log message The output dir for each individual job is now logged at DEBUG, instead of logging a string prefix common to all jobs that wasn't actually a directory or file by itself. | 15 November 2014, 07:14:32 UTC |
| c89cf72 | Dan LaRocque | 15 November 2014, 07:13:06 UTC | Adding a test single-map/no-reduce job This test will detect a regression on #806 / #826. I checked that it failed when I locally reverted 520e7c650e08ebc2d6b86bfa1515f032006259f4 (the commit that fixed this bug in manual testing). | 15 November 2014, 07:13:06 UTC |
| 0c28ed5 | Dan LaRocque | 15 November 2014, 07:08:46 UTC | Making HadoopPipeline.submit return Tool status This submit method is a thin wrapper around ToolRunner.run. ToolRunner.run returns the integer exit status produced by the Tool implementation it executes. Prior to this commit, submit returned void and threw the status away. Now the status is returned from submit. This change is groundwork for a new integration test method related to #806 and #826 that will submit a HadoopPipeline that submits a typical graphson-to-sequencefile dump job and needs to know whether the job succeeded. | 15 November 2014, 07:12:19 UTC |
| 35d2298 | Dan LaRocque | 15 November 2014, 06:18:11 UTC | HadoopXCompiler cleanups Commented code removal, import cleanups, unused variable deletion | 15 November 2014, 06:18:11 UTC |
| 520e7c6 | Dan LaRocque | 15 November 2014, 06:10:51 UTC | Set job output k-v along with Mapper output k-v Hadoop2Compiler (but not Hadoop1Compiler) uses ChainMapper and ChainReducer to string work together inside a single job. Hadoop2Compiler has an addMap method that creates a reducerless job. This method failed to set the job's output key and value classes, nor did ChainMapper.addMapper, so these jobs used the defaults that ship with Hadoop. These are almost always wrong for Titan-Hadoop jobs. Some output formats don't care about the configured k-v classes (e.g. TextOutputFormat), but SequenceFileOutputFormat definitely does. This change sets the job output k-v classes to match the Mapper k-v classes passed into Hadoop2Compiler.addMap. If a reducer is later added to the job, ChainReducer.setReducer overwrites the job output k-v, so I think this is safe. fixes #806 fixes #826 | 15 November 2014, 06:16:08 UTC |
| 9005483 | Dan LaRocque | 15 November 2014, 05:59:54 UTC | Hadoop compiler log additions & shared code reuse This extracts a common abstract superclass for the Hadoop1Compiler and Hadoop2Compiler subclasses. The abstract superclass is compiled under Hadoop 2 but contains only code with binary interop between both major versions, so it's in titan-hadoop-core. This commit adds more log messages to the Hadoop job compiler (including the new abstract base class). It logs when it creates jobs, configures input/output formats, and keys. Before executing any jobs, it also dumps the number of total jobs in a workload, then prints each each numbered job in the workload sequence along with the job name. When executing jobs, it prints the job number (in the workload sequence) and the name. That's all at INFO. With DEBUG enabled, input format, output format, key/value classes, and output paths are also logged. To make the above changes useful in the default configuration, this commit also modifies log4j-gremlin.properties to set the compiler class logger thresholds to ${gremlin.mr.log4j.level} (INFO by default). | 15 November 2014, 05:59:54 UTC |
| ee67f00 | Dan LaRocque | 14 November 2014, 23:15:52 UTC | Removed vestigial config keys from Hadoop1Compiler fixes #787 | 14 November 2014, 23:15:52 UTC |
| 5d6e3c5 | Dan LaRocque | 14 November 2014, 22:50:58 UTC | Make ES client version check defensive fixes #781 | 14 November 2014, 22:50:58 UTC |
| 736d050 | Dan LaRocque | 14 November 2014, 07:27:53 UTC | Adding six hour fork timeout to Surefire config This is mostly for our CI platform, which is capable of detecting hung builds based on the amount of time passed since their last line of output, but which nonsensically reliant on a nonstandard plugin to kill the jobs once flagged as hung. And the plugin only works on a manual bring-your-own-hardware install, not the CI-as-a-service cloud option. This is specifically for a strange deadlock in TransactionalGraphTestSuite (often line 484 when it deadlocks and BDB detects the cycle) that shows up in CI but not on my workstation. The root cause wasn't completely obvious at first read. | 14 November 2014, 07:27:53 UTC |
| 7243d05 | Dan LaRocque | 11 November 2014, 13:32:44 UTC | Use Titan SchemaContainer in Faunus's InputFormat Possible fix for #786. This could also make several calls to setSchemaProvider in the test code redundant. | 11 November 2014, 13:32:44 UTC |
| 94446d5 | Dan LaRocque | 05 November 2014, 20:35:33 UTC | Update RELEASING.md with reminder to update wiki (cherry picked from commit bceff85a1c83ac5084524f45313ffb81ebcb80ea) | 05 November 2014, 20:36:15 UTC |
| 6c7ae43 | Dan LaRocque | 28 October 2014, 03:58:26 UTC | Reduce severity of mundane HBase auth token stuff The two lines changed from ERROR to INFO in this commit never should have been at ERROR in the first place. | 28 October 2014, 03:58:26 UTC |
| d4432d7 | Dan LaRocque | 28 October 2014, 03:55:22 UTC | Merge branch 'titan05' of github.com:thinkaurelius/titan into titan05 | 28 October 2014, 03:55:22 UTC |
| 95ded2a | Dan LaRocque | 24 October 2014, 15:17:54 UTC | Sleep in tests until ES is ready ElasticsearchRunner.start now polls the ES logfile after starting the daemon in the background until a line ending in " started" appears in the log. This change is aimed at CI, where the hard-coded 2 second sleep wasn't long enough. The method polls every 500 ms for up to one minute, at which point it logs a message and returns. | 24 October 2014, 15:18:20 UTC |
| 4288691 | Dan LaRocque | 24 October 2014, 05:43:16 UTC | Add @Category for nondeterministic tests The BrittleTests category marks tests that are nondeterministic or intrinsically racy (e.g. hardcoded Thread.sleep without any synchronization primitive) and which fail intermittently due to timing problems on either CI, my workstation, or both. | 24 October 2014, 05:43:16 UTC |
| f4e9ac4 | Dan LaRocque | 24 October 2014, 05:42:11 UTC | Make ES bind to 127.0.0.1 for testing Checking to see if this resolves an issue with the ES tests on CI | 24 October 2014, 05:42:11 UTC |
| 162e806 | Dan LaRocque | 24 October 2014, 05:40:22 UTC | Update cassandra.yaml used in testing Replaced old cassandra.yaml template with one from 2.0.9. I did this after noticing that several timeout options honored by Cassandra 2.x were not present in our old cassandra.yaml template (and the short defaults may have caused a CI failure). | 24 October 2014, 05:41:24 UTC |
| 6c68f53 | Dan LaRocque | 24 October 2014, 00:20:12 UTC | Update gh-pages release script I'm modifying the old gh-pages-update.sh to work on the 0.5.x series. The AsciiDoc changes between 0.4 and 0.5 made it unsafe to use on 0.5, but now it just does javadoc and should be fine. Also updated the index.html template to match the current version live on the gh-pages website right now. | 24 October 2014, 01:31:08 UTC |
| 3dd7569 | Dan LaRocque | 15 October 2014, 17:23:17 UTC | Make Xdoclint:none activate only in JDK 1.8+ javadoc lint must be disabled on JDK 1.8 to prevent lint warnings from killing the build, but JDK 1.7 doesn't understand the -Xdoclint:none option that turns it off in 1.8. JDK 1.7 responds to this option by dying with nonzero status (which also kills the build). This commit adds a Maven profile activated on JDK version range "[1.8," that disables javadoc lint. | 15 October 2014, 17:23:17 UTC |
| de80743 | Dan LaRocque | 15 October 2014, 15:42:42 UTC | Merge branch 'titan05' of github.com:thinkaurelius/titan into titan05 | 15 October 2014, 15:42:42 UTC |
| 03a3b82 | Dan LaRocque | 15 October 2014, 14:56:27 UTC | Import cleanup in HadoopConfiguration The org.elasticsearch.common.collect.Iterators import never should have existed. This was almost certainly intended as a Guava Iterators import. | 15 October 2014, 14:56:27 UTC |
| d71100e | Dan LaRocque | 15 October 2014, 14:54:36 UTC | Catch Throwable in ReflectiveConfigOptionLoader For #785 | 15 October 2014, 14:54:36 UTC |
| 351bc0c | Matthias Broecheler | 10 October 2014, 19:41:15 UTC | Merge branch 'titan05' of https://github.com/thinkaurelius/titan into titan05 | 10 October 2014, 19:41:15 UTC |
| e78b59b | Matthias Broecheler | 10 October 2014, 19:41:07 UTC | Updated documentation. Added javadoc configuration to pom so that it compiles with java 8 | 10 October 2014, 19:41:07 UTC |
| 3086753 | Dan LaRocque | 10 October 2014, 02:48:19 UTC | Add 0.5.1 to compat manifest | 10 October 2014, 02:48:19 UTC |
| 9f6f5b9 | Dan LaRocque | 10 October 2014, 02:48:19 UTC | [maven-release-plugin] prepare for next development iteration | 10 October 2014, 02:48:19 UTC |
| 495402f | Dan LaRocque | 10 October 2014, 02:48:18 UTC | [maven-release-plugin] prepare release 0.5.1 | 10 October 2014, 02:48:18 UTC |
| ef725f7 | Dan LaRocque | 10 October 2014, 02:45:42 UTC | Updating titan.compatible.versions prop | 10 October 2014, 02:45:42 UTC |
| 2411f03 | Dan LaRocque | 10 October 2014, 02:39:42 UTC | Revert CommonsConfiguration to Enum.toString usage This commit effectively reverts 4c2c34226686ed91feb230af5c8a7a8e39b70403, but it also changes Timestamps.toString just delegate to the name method, so the strings stored in CommonsConfigurations backers will still be e.g. MILLI rather than Timestamps[ms]. Like 4c2c34226686ed91feb230af5c8a7a8e39b70403, this is orthogonal to KCVSConfiguration which uses a StandardSerializer to convert between enums instances and StaticBuffers. | 10 October 2014, 02:41:04 UTC |
| 579301b | Dan LaRocque | 10 October 2014, 00:57:15 UTC | Update upgrade instructions for 0.5.1 Copied @mbroecheler's UPGRADE.asc additions on 21478ebeea4f5df27b757373b3bf5dc046c34e29 to docs/upgrade.txt, added a date | 10 October 2014, 00:57:15 UTC |
| 1bc4136 | Dan LaRocque | 10 October 2014, 00:55:21 UTC | Update upgrade instructions for 0.5.1 | 10 October 2014, 00:55:21 UTC |
| 3426482 | Dan LaRocque | 10 October 2014, 00:53:53 UTC | Update changelog for 0.5.1 | 10 October 2014, 00:53:53 UTC |
| 66d86db | Dan LaRocque | 10 October 2014, 00:51:21 UTC | Update compat test to set index.search.directory This probably should not be GLOBAL_OFFLINE in the first place | 10 October 2014, 00:51:21 UTC |
| a6e751c | Dan LaRocque | 09 October 2014, 23:10:18 UTC | Merge branch 'titan05' of github.com:thinkaurelius/titan into titan05 | 09 October 2014, 23:10:18 UTC |
| 0333f4f | Dan LaRocque | 09 October 2014, 22:12:48 UTC | Add SolrIndex to reflective opt loader fast path Avoid triggering Reflections just to discover ConfigOptions defined on SolrIndex by including it in the set of hard-coded classes checked by ReflectiveConfigOptionLoader's `loadStandard` method | 09 October 2014, 22:12:48 UTC |
| 7e48baf | Dan LaRocque | 09 October 2014, 21:49:47 UTC | Reflections tweaks for 0.9.9 compatibility For #778 | 09 October 2014, 21:49:47 UTC |
