Log Parser Plugin

Hi ,
Can you also publish the compatibility information as the plugin did not work with the supported version of hudson (1.3.12)

I tried running this with the latest version of Hudson (1.355) but was not successful. Has anyone tried it yet?

Also, the format for the parsing rules file: Is it

_error /ERROR/_

or

error /ERROR/

(with or without the underscores?)

Awesome plugin, thanks!

Hi,

works fine for me on Hudson ver. 1.362 Very nice plugin.

The correct parsing rule is :

error  /ERROR/

And important point : default directory where for the parsing rule file is the directory which contains the jobs/ dir

Somethings wrong with the paths to the CSS-file in Hudson ver. 1.363.

eg. the link points to
/hudson/css/style.css

while it should be

/css/style.css

The directory from which the search for the parsing rule file begins is NOT always the directory which contains the "jobs" directory.

It is the current working directory of the process which starts the Hudson (only Windows platforms; other platforms i don't know).

Example:

1. Hudson is started through a batch file which is located in the directory "C:\Program Files\Hudson". Then this folder is the starting folder for the rule files.
2. Hudson is started as a windows service. Then the "%windir%\system32" or "%windir%\syswow64" folder is the starting folder for the rule files.

I think this is a bug of the plugin. The search should always start at "HUDSON_HOME". An other solution is that the plugin accepts environment variables in the file path for the rule files (this doesn't work at the moment too).

It would be nice to have the option to store parser rules in a text box from within the hudson GUI, rather than a file. I don't have file system access on our hudson server, but I do have admin access in the GUI.

EDITED:
I figured out how to copy a file to a specified location on the filesystem of the master hudson server using a maven plugin, however I am running into problems copying that file down to the slaves. I't like to copy the rule file to /tmp/minimalRules.txt but the "Copy to Slave" hudson plugin only puts the files into the root of $WORKSPACE. It doesn't look like this plugin supports reading environment variables so that doesn't help.

Hi all, nice plugin but Is there way how to mark more lines by some regex pattern(s)?? For example mark this lines of console output as error

[javac] Compiling 720 source files to /home/prcek/java_aisservis/GolemEJB/build/jar
[javac] /home/prcek/java_aisservis/GolemEJB/src/java/cz/aisservis/ejb/golem/utils/BeanMapper.java:4: ';' expected
[javac] package cz.aisservis.ejb.golem.utils
[javac] ^
[javac] 1 error

I have done it to modify plugin by myself but I will(I hope others as well) appreciate if it will in official release.

Thanks Tomas

I am seeing garbage characters (Hex characters) when the "Add timestamps" is selected in Job configure. Can anyone help me to overcome this issue or this is a bug in plugin?. Thanks in advance.

Here is how the output looks

ha:AAAAdB+LCAAAAAAAAABb85aBtbiIQSOjNKU4P0+vIKc0PTOvWK8kMze1uCQxtyC1SC8ExvbLL0llgABGJgZGLwaB3MycnMzi4My85FTXgvzkjIoiBimoScn5ecX5Oal6zhAaVS9DRQGQ1lnw9KIYADxiu4CBAAAA Submitting files for "Source" perforce...
ha:AAAAdB+LCAAAAAAAAABb85aBtbiIQSOjNKU4P0+vIKc0PTOvWK8kMze1uCQxtyC1SC8ExvbLL0llgABGJgZGLwaB3MycnMzi4My85FTXgvzkjIoiBimoScn5ecX5Oal6zhAaVS9DRQGQ1lnw9NokAMKzLyWBAAAA Warning: No files to submit for "Source"

Should be like this
23:46:44 Submitting files for "Source" perforce...
23:46:46 Warning: No files to submit for "Source"

A couple of other queries/suggestions. (Some may already be possible - if so, please let me know)

The first is cosmetic (but would stop distraction): Is it possible to configure the format of colours and fonts used? The orange on white for warnings is difficult to read, and the use of a non-uniformly-spaced font makes scanning the output cumbersome.

Re-parsing of logs: I would like to be able to modify the parser used and 'reparse' an existing log. There are times when you may find a false-positive, wish to change the parser or use a more verbose parser on the same log without having to re-run the build job.

Section grouping: Allow the expansion and reduction of lines for identified sections, similar to the top-level 'Error', 'Info' and 'Warning'. There are times where we have several hundred warnings or info messages and another level in the 'tree' would simplify their review.

Thoughts/comments gratefully received.

Thanks for the Plugin -- it's hugely useful (especially with Xcode builds which are very verbose and hard to spot errors in).  It's taken me a while to work out exactly how to get this running, because of the quirks in the way the path to the rules file is stored in the jobs, so for the benefit of other users here's how I got it running (the order of these steps is significant):

• Create a rules file on the hudson server
• Configure the hudson server (as in the Global Configuration section above), with the absolute path to where the rules file lives on the server
• Configure a job to use the log parser (as in the Job Configuration section above).  In the current version this will copy the location of the rules file into the job config.xml, so if you change the location of the rules file you will have to update the job scripts in a text editor

Hi,
Thanks for this wonderful plugin.

This plugin seems to be incompatible with java 1.5. I get this error when using on 1.5.
From the java docs, String supports isEmpty() only from 1.6 http://download.oracle.com/javase/6/docs/api/java/lang/String.html#isEmpty()

Can you pls make it compatible with 1.5?

16:09:50 FATAL: java/lang/String.isEmpty()Z
16:09:50 java.lang.NoSuchMethodError: java/lang/String.isEmpty()Z
16:09:50 at hudson.plugins.logparser.CompiledPatterns.setError(CompiledPatterns.java:18)
16:09:50 at hudson.plugins.logparser.LogParserUtils.compilePatterns(LogParserUtils.java:103)
16:09:50 at hudson.plugins.logparser.LogParserParser.<init>(LogParserParser.java:56)
16:09:50 at hudson.plugins.logparser.LogParserPublisher.perform(LogParserPublisher.java:52)
16:09:50 at hudson.tasks.BuildStepMonitor$1.perform(BuildStepMonitor.java:19)
16:09:50 at hudson.model.AbstractBuild$AbstractRunner.perform(AbstractBuild.java:603)
16:09:50 at hudson.model.AbstractBuild$AbstractRunner.performAllBuildSteps(AbstractBuild.java:582)
16:09:50 at hudson.model.AbstractBuild$AbstractRunner.performAllBuildSteps(AbstractBuild.java:560)
16:09:50 at hudson.model.Build$RunnerImpl.post2(Build.java:156)
16:09:50 at hudson.model.AbstractBuild$AbstractRunner.post(AbstractBuild.java:529)
16:09:50 at hudson.model.Run.run(Run.java:1361)
16:09:50 at hudson.model.FreeStyleBuild.run(FreeStyleBuild.java:46)
16:09:50 at hudson.model.ResourceController.execute(ResourceController.java:88)
16:09:50 at hudson.model.Executor.run(Executor.java:140)

This plugin doesn't work anymore with Hudson 1.398. "Console Output Parsing" menu doesn't appear on the System and Job Configuration Menus. It did work in Hudson 1.396.

This is a great plugin. It's amazing how long we trolled through our 200K+ line log files before. Of course, log files this large do stress some browsers beyond the breaking point, particularly when you try to navigate to a specific message. They don't crash, but they can become unresponsive.

My key feedback, though, is that the scanners seem to be case-sensitive. Thus, in order to get all of the interesting messages, I am trying expressions like this:

error /\[Ee\]\[Rr\]\[Rr\]\[Oo\]\[Rr\]/

Is there a way to mark the regular expression as being case-insensitive?