-
Bug
-
Resolution: Unresolved
-
Major
-
None
-
Jenkins 1.575 on Windows 7 32-bit; tap-plugin 1.8-1.20
I have some long existing Jenkins jobs with TAP tests configured. However at some update of either Jenkins or the tap-plugin the tap plugin stopped functioning correctly. This wasn't noticed earlier because the build jobs did not fail.
The plugin run ends with a message "Found matching files but did not find any TAP results"; in the workspace the testresult file is indeed empty. Analyzing this further it turns out that tap-plugin 1.7 works correctly, but any higher version seems to truncate the testresult file to 0 bytes before processing it.
Any ideas?
- depends on
-
-
- Open
-
[JENKINS-24250] TAP plugin truncates testresult.tap file
Hi,
I also have the same behavior on Windows Slave node (Win7 64bits / Java 1.8.0_101-b13):
I have a pretty long TAP result file (around 1800 tests representing 294887 characters/7290 lines in the file) and Only the 32 firsts are parsed, and the last one is truncated. After multiple tests, it seems that the file is always truncated at the same position: only the first 4812 characters are read.
But I see the whole file in Console Output.
Maybe the file is read before it has been completely written ?
Here is my result file: Result.tap
EDIT: I'm running Jenkins 2.7.2
Regards, Amaury LAURENT.
Amaury LAURENT
added a comment - - edited Hi,
I also have the same behavior on Windows Slave node (Win7 64bits / Java 1.8.0_101-b13):
I have a pretty long TAP result file (around 1800 tests representing 294887 characters/7290 lines in the file) and Only the 32 firsts are parsed, and the last one is truncated. After multiple tests, it seems that the file is always truncated at the same position: only the first 4812 characters are read.
But I see the whole file in Console Output.
Maybe the file is read before it has been completely written ?
Here is my result file: Result.tap
EDIT: I'm running Jenkins 2.7.2
Regards, Amaury LAURENT. Hi, I believe that once https://issues.jenkins-ci.org/browse/JENKINS-19647 has been fixed, this issue will disappear.
IIRC, one of the performance problems is caused by the plug-in copying loads of files from the slave to the master.
That's very likely to be what's causing this issue. Will link both issues here in JIRA.
I have discovered something similar from my Windows build slaves; some of the generated .tap files get truncated somewhere between the build slave's workspace and the Jenkins master, although in my case the truncation happens part-way through the file, cutting off the last few test results. I'm currently running tap-plugin 1.18 on Jenkins 1.552, and this only seems to happen on build slaves running on Windows 7 with Oracle Java 1.7.0.