Investigating our issue further, it seems that it is a different problem. The project configuration is such that it executes some sub projects. What we see in the console output (with the timestamper plugin enabled) is that the next queued job starts to run a few seconds after the first one has started its (blocking) sub job. File a new issue?

Issue seems to be unresolved in 1.606
Environment is CentOS 6.6 slave and git version 1.8.3.1

o_ozardic you will need to provide much more information about the conditions which show the problem. I doubt a developer will be persuaded to spend significant time trying to guess the conditions under which you have seen the problem. Even if they do spend that time guessing more about the problem, they may guess incorrectly and fix something that is not the issue you've seen.

Please provide step by step instructions which show the problem, preferably on a clean installation of a recent Jenkins version (either the current long term support version or the most recent weekly version). You may also help by installing the "support" plugin and including the information from that plugin. Or, at least include the versions of plugins you're using in addition to the command line git version you have installed.

OS: Ubuntu 12.04 64-bit
Jenkins ver: 1.609.1 (upgraded from 1.580.3)
disk-usage plugin: 0.25 (upgraded from 0.24)
Matrix Project Plugin: 1.6 (upgraded from 1.4.1)

I didn't have this issue when I was using ver 1.580.3.
After upgrading Jenkins ver 1.609.1 (together with other Jenkins plugins update),
I started to see this issue.

My jenkins job is using multi-config project type (aka Matrix project),
and has 3 variants.
I didn't set to be run concurrently,
so the 3 variants will start the build one by one.

In the previous version 1.580.3,
it will reuse the same workspace folder for all of 3 variant builds.

In the new version 1.609.1,
when multiple jobs are triggerd,
I can see sometimes the first variant build of the second job
will start to create and use workspace@2
instead of re-using original worksapce.

Note:
I'm also using disk-usage plugin.
At the end of build,
it will trigger that plugin to calculate workspace disk space.
My work space tends to be large and sometimes it take several minutes to finish.

I suspect it's that disk-usage plugin still occupying the original worksapce,
but the second job in queue starts earlier,
so it creates workspace@2 for the first variable build.

From workspace@2,
I always see only one variant only under AXIS folder.