Bug: Integration test build for blitz has configuration troubles

[mtbcarroll]

With a fresh build from current develop,

./build.py -f components/blitz/build.xml integration

fails with,

org.springframework.beans.TypeMismatchException: Failed to convert property value of type 'java.lang.String' to required type 'int' for property 'maxPoolSize'; nested exception is java.lang.NumberFormatException: For input string: "${omero.db.poolsize}"

If into the suite macrodef in components/antlib/resources/lifecycle.xml one inserts,

<jvmarg value="-Domero.db.poolsize=${omero.db.poolsize}"/>

then the -verbose argument confirms that -Domero.db.poolsize=10 is passed but the NumberFormatException exception remains unchanged.

Annoyingly, -verbose also reveals that -Domero.data.dir=/OMERO/ is passed, which I bet causes some other bean creation to fail, because I don't have that directory, instead I have,

$ omero config get omero.data.dir
/Users/mark/var/omero/

[jamoore]

​https://github.com/openmicroscopy/openmicroscopy/pull/1000

Integration tests are not yet passing, but the omero.db.poolsize issue is taken care of. We'll need to decide how far we want to go with respect to getting that particular ant target passing as a part of PR 1000.

[Josh Moore <josh@…>]

(In [0a7029ca7690c4cd1d3c4b64e38377f8a2ec24ba/ome.git] on branch develop) Merge pull request #1000 from joshmoore/10560-blitz-itest

Initial clean up of blitz integration tests (Fix #10560)

[mtbcarroll]

Should I file a separate ticket about integration tests failing miserably when /OMERO/ doesn't exist regardless of omero.data.dir's setting?

[jamoore]

You can, or you can re-open this. I locally have omero.data.dir set for the tests, but that should happen far more automatically so that the tests work for anyone. (And importantly, so they don't overwrite any files,say, on gretzky)

[mtbcarroll]

The outstanding issue is that the Blitz integration tests require /OMERO/ to exist whatever omero.data.dir's configuration.

[jamoore]

Where have you tried setting it, Mark?

[mtbcarroll]

omero config set omero.data.dir <whatever> before starting the server -- is that not the way to do it? Perhaps the developer docs on tests need adjustment.

[jamoore]

The Java code knows nothing about the etc/grid/config.xml. That's processed solely by Python and icegridnode to pass a properties file to the Java processes. The two options which are possibly open to the server Java developer are: etc/local.properties.example (which yes, should be named default) and the ICE_CONFIG environment variable. (In Eclipse, I pass the values directly to the JVM via the "Installed JREs" dialog)

[jburel]

Moving the reopened ticket to 4.3 sprint

[jburel]

Due to the time required to identify the issue, moving it back to 4.x

[mtbcarroll]

On dev_4_4 ./build.py -f components/server/build.xml test presently gives me,

Failed to convert property value of type 'java.lang.String' to required type 'int' for property 'maxPoolSize'; nested exception is java.lang.NumberFormatException: For input string: "${omero.db.poolsize}"

(Update: ah, yes, now I see PR 1000 is still in ​http://hudson.openmicroscopy.org.uk/view/Mgmt/job/SCC-unrebased-prs/lastSuccessfulBuild/console)

[jamoore]

Rebased: ​https://github.com/openmicroscopy/openmicroscopy/pull/1502

@mtbc: if you think that covers this ticket, can you please re-close?

[mtbcarroll]

Blazej's improvements to developers/testing.txt address the outstanding issues here.

[Josh Moore <josh@…>]

(In [216438686cd1245e90babd82142b31fc836d3ae7/ome.git]on branches master, dev_4_4) Merge pull request #1502 from joshmoore/rebased/dev_4_4/10560-blitz-itest

Initial clean up of blitz integration tests (Fix #10560) (rebased onto dev_4_4)