This message was deleted.
# performance
s
This message was deleted.
c
Gradle 8 release notes list the changes.
p
Gradle 8 requires a bit more memory. The metaspace is especially concerned. The first thing to do would be to check if your build daemon is not spending too much time doing GC.
t
The metaspace is especially concerned.
Could you elaborate a little on this? I have currently pending investigation regarding OOM in Metaspace. Project itself runs on Gradle 8 and I don't have a clue what could be wrong on KGP side looking at provided heap dump.
c
^^^ Gradle 8 uses more metaspace. If you are hitting on OOM there increase the metaspace size.
t
in case of issue reporter it didn't help. It just delayed it.
c
in that case it’s either a leak or the increase was not sufficient.
m
I will check whether gc is involved and let you know. What would be the easiest way to do it besides gradle scan? perhaps gradle doctor could tell?
t
I think increase to 7g should be sufficient πŸ™‚ Though in this case Gradle daemon not died immediately with OOM:Metaspace, but showed
The Daemon will expire after the build after running out of JVM Metaspace.
and then stopped itself
c
are we talking about 7g of metaspace, or 7g heap?
t
metaspace. Heap was always ok and not hitting the limit
πŸ‘ 1
c
typically in that case (what appears to be a leak) is to grab a heap dump and analyze that in a memory profiler to figure out what is being held on to. and then why that is the case.
for metaspace, the culprits are often dynamically generated classes - proxies or otherwise.
t
issue was in android databinding and not related to Gradle
πŸ™Œ 1
d
@Tapchicoma Can you please link an issue about databinding if it exists? I want to know about possible problems with databinding regarding ram usage
t
@Dmitriy Voronin this one
d
thanks!