[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
Re: [tycho-user] What could have triggered the change of the caching behavior of "Adding repository" ?
|
Actually, the VM run out of heap. After increasing Xmx from 128m to 192m the build succeeded (yes.. my server is weak).
I asked for soft or weak-caching, not because I thought there is the problem, but to explain myself, what caused the switch from "Adding
repository (cached)" to "Adding repository" at exactly the same time in build? After increasing the heap, Tycho performs
"Adding
repository (cached)" again. This behaviour could be explained with a memory sensitive cache?
Thanks,
Bernd
2012/3/4 Igor Fedorenko
<igor@xxxxxxxxxxxxxx>
As far as I can tell, you have the same repository referenced from
multiple reactor modules. Parsing repository metadata is rather
expansive, so Tycho reads it once and then reuses the same repository
metadata for all modules that reference the repository. Whether this is
a weak/soft or a hard reference does not matter because complete
metadata needs to be read in memory and your JVM does not have enough
heap for that. You need to either increase heap limit or reduce amount
of metadata (by referencing specific 3.7.x build instead of all
versions, for example).
--
Regards,
Igor
On 12-03-04 12:50 PM, Bernd Vogt wrote:
Most probably it's like you say... my build server is not the strongest
one and I have to keep memory consumption very low.
Do you have any ideas, what caused Tycho the switch from "Adding
repository (cached)" to "Adding repository"? Does Tycho make use of soft
or weak caching at this point?
2012/3/4 Igor Fedorenko <igor@xxxxxxxxxxxxxx <mailto:igor@xxxxxxxxxxxxxx>>
Most likely your build already was on the threshold of running out of
heap. New eclipse version (i.e. 3.7.2) brought additional p2 metadata
and pushed the build over the available heap limit.
--
Regards,
Igor
On 12-03-04 10:32 AM, Bernd Vogt wrote:
Hi,
from one day to the other one of my build jobs begun to fail
because of
java.lang.OutOfMemoryError: Java heap space. And I'm still trying to
figure out why?
I noticed a little behavioral change while the tycho-surefire-plugin
tries to resolve the org.eclipse.equinox.launcher from my locally
mirrored Indigo repository. Befor the build started to fail, the
Indo
repository was already cached by Tycho, but now Tycho decides to
add a
"fresh" repository, see snippets below (Adding repository
(cached) vs.
Adding repository).
What could have triggered this behavioral change?
Neither the build configuration, hardware or software was
changed nor
the source code. BUT around this time I have updated the local
Indigo
mirror with the new 3.7.2 stuff.
Thanks in advance
Bernd
The good old days
<http://jenkins.sourcepit.org/job/b2_nightly/213/consoleFull>__):
[INFO] --- tycho-surefire-plugin:0.14.1:__test (default-test) @
org.sourcepit.b2.examples.__simple.module.tests ---
[...]
[INFO] Adding repository *(cached)
*http://p2.sourcepit.org/__eclipse/indigo/
(http://jenkins.sourcepit.org/__job/b2_nightly/209/consoleFull
<http://jenkins.sourcepit.org/job/b2_nightly/209/consoleFull>__):
[INFO] --- tycho-surefire-plugin:0.14.1:__test (default-test) @
org.sourcepit.b2.examples.__simple.module.tests ---
[...]
[INFO] Adding repository
http://p2.sourcepit.org/__eclipse/indigo/
_________________________________________________
tycho-user mailing list
tycho-user@xxxxxxxxxxx <mailto:tycho-user@xxxxxxxxxxx>
https://dev.eclipse.org/__mailman/listinfo/tycho-user
<https://dev.eclipse.org/mailman/listinfo/tycho-user>
_________________________________________________
tycho-user mailing list
tycho-user@xxxxxxxxxxx <mailto:tycho-user@xxxxxxxxxxx>
https://dev.eclipse.org/__mailman/listinfo/tycho-user
