Sure, easy enough to do! Can you please raise a JIRA issue for this usecase?
Cheers > On 1 Apr 2015, at 23:56, tran, phong <phong.t...@emc.com> wrote: > > Hi Mauro, > Yes we’re running in multi-threading mode with 10 threads for concurrent > story execution. > > EmbedderControls[batch=false,skip=false,generateViewAfterStories=true,ignoreFailureInStories=true,ignoreFailureInView=false,verboseFailures=true,verboseFiltering=false,storyTimeoutInSecs=108000,failOnStoryTimeout=false,threads=10] > > We’re trying to speed up the test execution (Selenium-based tests) with > multiple threads (threads=10) targeting a Selenium Grid with 10 nodes. > > I noticed that the reported duration for each story is also high and that is > why the sum of all story durations added up to a high total number. Does this > mean the duration at the story level has been affected with multi-threading > execution mode? Do we need to de-normalize (divided by 10 in this case) the > duration for each story to get its accurate execution time? > > It would be great if the de-normalization for story duration can be at > addressed at the JBehave framework level to handle the multi-threading > execution mode. > > Thanks, > Phong > > From: Mauro Talevi [mailto:mauro.tal...@aquilonia.org] > Sent: Saturday, March 28, 2015 1:56 AM > To: user@jbehave.codehaus.org > Subject: Re: [jbehave-user] Story duration are not calculated correctly in > multi-thread execution > > Hi Phong, > > I would imagine you're running in multi-threading mode. The total is > calculated as simply be the sum of each duration. If you have a single > thread then it will coincide with the build duration (more or less), but you > use multiple threads it'll be off by a factor which should correspond to the > number of threads. > > Arguably we could use the information of the threads to renormalise the total > by the number of threads. This formula would still work in the case of a > single thread. > > Can you confirm that you're using multiple threads? > > Cheers > > On 25/03/2015 08:22, tran, phong wrote: > Hi Mauro, > Thanks a lot for updating the core example template! We can access and > integrate the durations from storyDurations.props into our custom HTML report > now. However, we just notice another issue. The total duration reported in > storyDurations.props for all stories is much higher (8 times) that what we > usually see in our test run. Basically we have a configuration in TeamCity to > run our Selenium-based tests and the test run took 5 hours and 25 minutes as > reported by TeamCity. This is the normal time range we expected for our test > run. > > > > Results > Started > Changes > Started > Duration > > #177 > <image001.png> NGIS #: 6234 Beacon #: 3224 Pass Rate: 83.43949% passed: 262 > failed: 52 pending: 51 > <image002.png> View > Changes (7) > 24 Mar 15 09:55 > 5h:25m > Beacon_Selenium_W2K8R2_13 > None > > > However, the total duration aggregated for all stories as reported by JBehave > is unusually high (more than 48 hours). This duration is not correct! > Reports Overview > > Stories > Scenarios > Steps > Download as CSV > Total > Excluded > Total > Successful > Pending > Failed > Excluded > Total > Successful > Pending > Failed > Not Performed > Ignorable > Duration (hh:mm:ss.sss) > 279 > 0 > 345 > 210 > 82 > 53 > 0 > 9,195 > 3,751 > 230 > 66 > 3,699 > 1,449 > 48:39:21.000 > > storyDurations.props: > #org.jbehave.core.embedder.StoryManager > #Tue Mar 24 15:20:14 PDT 2015 > total=175161000 > > Do you have any idea what could be causing the much higher story duration > reported in storyDurations.props? Is there any JBehave configuration I should > check in our setup? > > Thanks, > Phong > > > > From: Mauro Talevi [mailto:mauro.tal...@aquilonia.org] > Sent: Saturday, March 21, 2015 7:00 AM > To: user@jbehave.codehaus.org > Subject: Re: [jbehave-user] Story duration are not calculated correctly in > multi-thread execution > > Hi Phong, > > the core example template was out-of-date. I've pushed the updated one. > > You'll see that the storyDurations are taken from a different template object > called "storyDurations". > > There is a case for integrating this info into the Report and/or ReportTable > objects. Please raise a JIRA for this against version 4.x. > > FYI, these objects are instantiated in the TemplateableViewGenerator. > > Cheers > > On 20/03/2015 21:50, tran, phong wrote: > Hi Mauro, > For example, take a look at the custom report packaged in JBehave core > example (/behave-3.9.5/examples/core/src/main/java/ftl/custom-reports.ftl). > In this FreeMarker template, a data object reportsTable is accessed and used > to populate the content for HTML reports. > > <#assign reportNames = reportsTable.getReportNames()> > > Through reportsTable data object, other stats information (including story > duration) is also accessed. > > <#assign stats = report.getStats()> > <@renderMillis stats "duration"/> > > Questions: > > 1. Where is reportsTable data model coming from? How is it created and > injected into this FreeMarker template file (.ftl)? > > 2. As the story duration is no longer available in the stats object > (story .stats file) with JBehave 3.9.4+ > (https://jira.codehaus.org/browse/JBEHAVE-1041), and if I want to use and > access the story duration available in storyDurations.props file from the > same FreeMarker template file, how can I do this? > > Thanks, > Phong > > From: Mauro Talevi [mailto:mauro.tal...@aquilonia.org] > Sent: Friday, March 20, 2015 2:16 AM > To: user@jbehave.codehaus.org > Subject: Re: [jbehave-user] Story duration are not calculated correctly in > multi-thread execution > > Not sure what you mean. Where do you want to make available the story > durations? > > On 20/03/2015 00:17, tran, phong wrote: > Thanks Mauro and Brent for confirming the fix. It’s actually my fault. I did > not update all other modules in the project with the latest JBehave 3.9.5. I > could see the storyDurations.props file created with expected story duration > time now. My next question is how I can take the information in > storyDurations.props file and make them available for access in the > FreeMarker templates (.ftl) to produce the HTML views? > > Thanks, > Phong > > From: Mauro Talevi [mailto:mauro.tal...@aquilonia.org] > Sent: Thursday, March 19, 2015 3:00 AM > To: user@jbehave.codehaus.org > Subject: Re: [jbehave-user] Story duration are not calculated correctly in > multi-thread execution > > Can you provide a sample project reproducting this behaviour. > > This problem should be fixed in latest release. > > On 17/03/2015 22:39, tran, phong wrote: > When I ran stories with a single thread (serial), the story duration in > reports looked correct. The duration time is in minutes as expected. See the > attached report (single-thread-reports.png). However, when stories are > executed with multiple threads (currently), the story duration is not > calculated correctly (See Multiple-Thread-Reports.png). The duration time is > in in sub-seconds and is not what expected. I’m using JBehave 3.9.2 and > thought this problem could be related to > https://jira.codehaus.org/browse/JBEHAVE-1041 but upgrading to the latest > JBehave 3.9.5 does not seem to help with this problem. > > Here is the code fragment for building story report. > > .useStoryReporterBuilder(new StoryReporterBuilder() > .withDefaultFormats() > .withFormats(CONSOLE, TXT, HTML_WEB) > .withFailureTrace(true) > .withFailureTraceCompression(false) > .withViewResources(viewResources) > .withCrossReference(xref) > ); > > Does anyone have an idea what’s could be the problem here? > > Thanks in advance, > Phong > > > > > > > > > > --------------------------------------------------------------------- > To unsubscribe from this list, please visit: > > http://xircles.codehaus.org/manage_email > > > >
