Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | Merge pull request #6223 from jefimm/patch-2 | Jon Bratseth | 2018-06-18 | 1 | -1/+1 |
|\ | | | | | make function public | ||||
| * | make function public | Jefim Matskin | 2018-06-18 | 1 | -1/+1 |
|/ | | | make function public to facilitate other lingustics | ||||
* | Merge pull request #6219 from vespa-engine/balder/use-single-precomputed-hashmap | Henning Baldersheim | 2018-06-18 | 15 | -112/+133 |
|\ | | | | | Precompute and store attribute and executor id together in a single h… | ||||
| * | Add missing ')' | Henning Baldersheim | 2018-06-18 | 1 | -1/+1 |
| | | |||||
| * | Add test for getNamePrefix | Henning Baldersheim | 2018-06-18 | 1 | -2/+19 |
| | | |||||
| * | Use correct prefix name when selecting executor. | Henning Baldersheim | 2018-06-17 | 3 | -9/+4 |
| | | |||||
| * | - Remove the execute(string, ...) and force the use of ExecutorId. | Henning Baldersheim | 2018-06-17 | 10 | -75/+59 |
| | | | | | | | | | | - Remove some double bookkeeping in AttributeWriter. - Ensure that we always use attribute.getNamePrefix() to compute executor id. | ||||
| * | Rewrite test to enure AttributeWriter is given a fully populated ↵ | Henning Baldersheim | 2018-06-17 | 2 | -26/+31 |
| | | | | | | | | AttributeManager | ||||
| * | Add getNamePrefix to get the prefix before any '.' | Henning Baldersheim | 2018-06-17 | 1 | -0/+5 |
| | | |||||
| * | Choose executorId the same way both places. | Henning Baldersheim | 2018-06-16 | 1 | -3/+7 |
| | | |||||
| * | Precompute and store attribute and executor id together in a single hashmap. | Henning Baldersheim | 2018-06-16 | 2 | -7/+18 |
| | | | | | | | | the executor id so complex fields use the same executor. | ||||
* | | Merge pull request #6220 from jefimm/patch-1 | Jon Bratseth | 2018-06-18 | 1 | -1/+1 |
|\ \ | | | | | | | Please make the class public | ||||
| * | | Please make the class public | Jefim Matskin | 2018-06-18 | 1 | -1/+1 |
|/ / | | | | | Please make the class public to facilitate work on custom liguistics | ||||
* | | Merge pull request #6216 from vespa-engine/arnej/render-maps-in-json | Arne H Juul | 2018-06-18 | 2 | -4/+81 |
|\ \ | |/ |/| | Arnej/render maps in json | ||||
| * | prefer handling Inspectable before JsonProducer | Arne Juul | 2018-06-15 | 1 | -2/+2 |
| | | | | | | | | | | | | * ensure we can handle special cases in renderInspector() instead of delegating to default rendering in JsonProducer interface. | ||||
| * | wrap map in object instead of rendering to intermediate string | Arne Juul | 2018-06-15 | 1 | -32/+23 |
| | | |||||
| * | render maps with string keys as JSON objects | Arne Juul | 2018-06-15 | 2 | -4/+90 |
| | | | | | | | | | | | | | | | | * if structured data is an array of <key,value> objects, we prefer to render it as a JSON object since it is most likely a map in the document definition. * makes docsum rendering a closer match with document rendering and feed input format. | ||||
* | | Merge pull request #6217 from vespa-engine/balder/multiple-threads-in-mbus | Henning Baldersheim | 2018-06-15 | 8 | -24/+84 |
|\ \ | | | | | | | Balder/multiple threads in mbus | ||||
| * | | Use std::max, and do it on the correct place. | Henning Baldersheim | 2018-06-15 | 2 | -2/+2 |
| | | | |||||
| * | | Control threadpool and dispatch of encode/decode by config. | Henning Baldersheim | 2018-06-15 | 7 | -18/+70 |
| | | | |||||
| * | | Use multiple threads for handling requests, if allowed. | Henning Baldersheim | 2018-06-15 | 2 | -8/+16 |
| | | | |||||
* | | | Merge pull request #6215 from vespa-engine/hakonhall/warn-on-timeout | Håkon Hallingstad | 2018-06-15 | 1 | -1/+1 |
|\ \ \ | | | | | | | | | Warn on timeout | ||||
| * | | | Warn on timeout | Håkon Hallingstad | 2018-06-15 | 1 | -1/+1 |
| |/ / | |||||
* | | | Merge pull request #6201 from vespa-engine/mpolden/back-off-failed-job-retry | Martin Polden | 2018-06-15 | 11 | -261/+460 |
|\ \ \ | | | | | | | | | Back off when retrying failing jobs | ||||
| * | | | Always run job if targets change | Martin Polden | 2018-06-15 | 4 | -52/+107 |
| | | | | |||||
| * | | | Back off when retrying failing jobs | Martin Polden | 2018-06-14 | 7 | -35/+89 |
| | | | | |||||
| * | | | Move deployment spec helper methods to DeploymentSteps | Martin Polden | 2018-06-14 | 4 | -74/+103 |
| | | | | |||||
| * | | | Rename DeploymentOrder -> DeploymentSteps | Martin Polden | 2018-06-13 | 4 | -26/+30 |
| | | | | |||||
| * | | | Pull up Versions class | Martin Polden | 2018-06-13 | 2 | -72/+129 |
| | | | | |||||
* | | | | Merge pull request #6190 from vespa-engine/lesters/random-normal-match | Geir Storli | 2018-06-15 | 9 | -46/+275 |
|\ \ \ \ | | | | | | | | | | | Add randomNormal.match | ||||
| * | | | | Use correct seed and don't use spare for randomNormalStable | Lester Solbakken | 2018-06-15 | 3 | -13/+16 |
| | | | | | |||||
| * | | | | Rename feature randomNormalMatch -> randomNormalStable | Lester Solbakken | 2018-06-15 | 6 | -37/+37 |
| | | | | | |||||
| * | | | | Move match version of random normal to it's own feature | Lester Solbakken | 2018-06-14 | 9 | -57/+217 |
| | | | | | |||||
| * | | | | Remove unnecessary logging in unit test | Lester Solbakken | 2018-06-14 | 1 | -2/+0 |
| | | | | | |||||
| * | | | | Fix random normal match test | Lester Solbakken | 2018-06-14 | 2 | -8/+31 |
| | | | | | |||||
| * | | | | Refactor out normal distributed random generator | Lester Solbakken | 2018-06-14 | 3 | -48/+65 |
| | | | | | |||||
| * | | | | Pass random generator by reference | Lester Solbakken | 2018-06-14 | 1 | -1/+1 |
| | | | | | |||||
| * | | | | Add match to output in randomNormal test | Lester Solbakken | 2018-06-13 | 1 | -1/+1 |
| | | | | | |||||
| * | | | | Add randomNormal.match | Lester Solbakken | 2018-06-13 | 3 | -13/+41 |
| | | | | | |||||
* | | | | | Merge pull request #6206 from vespa-engine/bjorncs/zts-client | Bjørn Christian Seime | 2018-06-15 | 21 | -154/+386 |
|\ \ \ \ \ | |_|_|/ / |/| | | | | Bjorncs/zts client | ||||
| * | | | | Store CA certificates to disk | Bjørn Christian Seime | 2018-06-15 | 1 | -14/+35 |
| | | | | | |||||
| * | | | | Add temporary workaround for missing path in zts uri | Bjørn Christian Seime | 2018-06-15 | 1 | -1/+1 |
| | | | | | |||||
| * | | | | Replace use of AthenzCertificateClient with DefaultZtsClient | Bjørn Christian Seime | 2018-06-15 | 3 | -83/+28 |
| | | | | | |||||
| * | | | | Add method to ZtsClient to retrieve identity certificate | Bjørn Christian Seime | 2018-06-15 | 18 | -57/+323 |
| | |_|/ | |/| | | |||||
* | | | | Merge pull request #6214 from ↵ | Henning Baldersheim | 2018-06-15 | 2 | -76/+67 |
|\ \ \ \ | |/ / / |/| | | | | | | | | | | | vespa-engine/balder/avoid-redeserialization-of-documentid Only deserialize the docId on the initial deserialize of the header. | ||||
| * | | | Reuse common code | Henning Baldersheim | 2018-06-15 | 1 | -4/+2 |
| | | | | |||||
| * | | | Only deserialize the docId on the initial deserialize of the header. | Henning Baldersheim | 2018-06-14 | 2 | -76/+69 |
| | | | | | | | | | | | | | | | | Just skip and verify on subsequent invocations. | ||||
* | | | | Merge pull request #6209 from ↵ | Jon Bratseth | 2018-06-15 | 1 | -11/+11 |
|\ \ \ \ | |/ / / |/| | | | | | | | | | | | vespa-engine/hakonhall/only-count-time-with-no-threads-available Only count time with no threads available | ||||
| * | | | Change field name | Jon Bratseth | 2018-06-15 | 1 | -3/+3 |
| | | | | |||||
| * | | | Only count time with no threads available | Håkon Hallingstad | 2018-06-14 | 1 | -9/+9 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Found this edge-case with logAndDie in ThreadPoolProvider: - A Runnable executes and finishes at time T0 - No Runnables execute for a long time t > maxThreadExecutionTimeSeconds - At time T1 = T0 + t, a bursts of Runnables N > maxthreads arrives such that the (maxthreads+1) Runnable will be tried executed and rejected before any of the first maxthreads Runnables finish. - Then, logOrDie will be called immediately. That is, a sufficiently silent period followed by a burst of Runnables cause the process to die. This PR will make sure that the process dies only if maxThreadExecutionTimeSeconds seconds pass with no available threads, like the documentation says. I found this as I was looking for reasons why the Cluster Controller have had logAndDie problems. The CC has maxthreads of 10, and was getting bursts of 6-8 of a particular request type (reduced in PR 6207). |