Daniel Erat
11:17 AM Today

this seems much more important to me than adding alerts about the frequency of chrome crashes, and should probably be the top priority here, right? sometimes chrome is going to be very crashy. the CQ needs to avoid overloading the network infrastructure when artifacts are larger than expected.


Ningning Xia
12:11 PM Today
each core dump file is ~ 300Mb. What will be the good limit size?

	
Ningning Xia
12:11 PM Today
maybe we need a configurable flag to turn on/off the offloading of dump files when this kind of crisis happens again?

	
Daniel Erat
12:19 PM Today
what about if we just offload at most one core dump (while logging the existence of others)? i'm guessing/hoping that the normal state is for there to be zero chrome crashes, so if we have multiple dumps, it might be safe to assume that chrome is in a crash loop.

would only offloading one dump have prevented the meltdown, or would we still have been in trouble due to the number of bots trying to offload?
	
Ningning Xia
2:44 PM Today
I'm not sure how easy it is to identify whether crashes are same and only upload one if there're duplicated. maybe symbolized stacks as James Cook mentioned above will help