Closed zzambers closed 9 months ago
This is done as part of effort to reduce disk space usage by test jobs on our infra. Jtreg and tck jobs are both in top space consumers. Compression ratio of xz
is much better than gz
for jtreg/tck results.
pls, add some unittests. Lib is enough. I'm unsure xml.xz is anything..Do you know? The plugin depends on lib, is it realy necessary to duplicate the dependende?
pls, add some unittests. Lib is enough. I'm unsure xml.xz is anything..Do you know? The plugin depends on lib, is it realy necessary to duplicate the dependende?
.xml.xz
is for tck results. These are big xmls unlike Jtregs where there are many small ones in archive. We currently use .xml.gz
for tcks.thanx for explanations. Pls add those unittests. For all three patched cases.
I have extended existing tests to also test xz compressed results.
thanx! The boiler plate to create the tmp file is terrible, but nvm, no more torturing of you :)
jjust for record, chartjs-1-rebasedToRefactoredReleased24 successfully reabsed
@judovana thank you
@RadekCap @zzambers I had released jtreg plugin 2.5, as we wil be updating our jenkins veryu soon, the 2.5 should make it in, and thus we will be able to use tar.xz immediately.
This adds support for
xz
compressed results of jtregs and tcks. Tested on artifacts from jtreg and jdk jobs repacked with xz compression in local jenkins instance. Worked for me.Notes:
JtregReportParser.java
already had code for .tar.xz, usingXZCompressorInputStream
fromcommons-compress
. However it fails for .tar.xz with ClassNotFoundException (org.tukaani.xz.XZInputStream). commons-compress has optional dependency on org.tukaani.xz, which is used in XZCompressorInputStream. This did not work for me even when org.tukaani xz dependency was added, so I switched directly toXZInputStream
fromorg.tukaani.xz
, which did work.