To close this thread, the disconnection of linux agents was a side effect
of the kernel settings which were using a too long tcp timeout
Gavin applied the settings recommended in this doc (
https://support.cloudbees.com/hc/en-us/articles/115001369667-Dedicated-SSH-agent-gets-disconnected
) and it
As explained over slack it moves the problem and creates by design new
ones, if urging a revert is easier IMHO.
Long term fix can be a ref counting or so but no regression, at least in
single thread, is better anyway for end users I think.
Le lun. 23 août 2021 à 12:12, Falko Modler a écrit :
>
There is an open PR for that as well: https://github.com/apache/maven/pull/521
-
To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org
For additional commands, e-mail: dev-h...@maven.apache.org
The one Olivier spotted is not fixed.
Le lun. 23 août 2021 à 11:02, Falko Modler a écrit :
> Why revert? Are you aware of any other regression caused by that?
> Both known regressions have been fixed (not yet merged).
>
> Having said that, as discussed previously, a solution without a TL is
>
Why revert? Are you aware of any other regression caused by that?
Both known regressions have been fixed (not yet merged).
Having said that, as discussed previously, a solution without a TL is
preferable in the long run but probably requires a more invasive approach that
is too big for a 3.8.x
+1 to revert this threadlocal thing in a 3.8.3
Breaks too silently mojo and is hard to track and technically it is
solvable on our side.
Le lun. 23 août 2021 à 08:29, Slawomir Jaranowski
a écrit :
> Hi,
>
> I think that https://issues.apache.org/jira/browse/MNG-7215 is also
> important, it
Hi,
I think that https://issues.apache.org/jira/browse/MNG-7215 is also
important, it breaks the build maven site for projects ...
Please don't extend the manual process ... nowadays we have everything to
do automatically, every (or most of) plugin has IT tests ... so we need
just to run its