> CASSANDRA_HEAPDUMP_DIR Is for JVM out of memory. You were seeing the OS kill the JVM because of low os memory.
Cheers ----------------- Aaron Morton Freelance Developer @aaronmorton http://www.thelastpickle.com On 22/08/2012, at 4:28 PM, Xu Renjie <xrjxrjxrj...@gmail.com> wrote: > Guille, > Thanks for your reply. I seem to find where is the problem. I guess it is > because memory is used up not the jvm heap( I use micro ec2 instance just for > toy use). Seems that I have set the cassandra heap too high to avoid gc to > effect my performance, but the total memory is not enough. So I have set a > smaller cache value for my application. Currently, it works well. > > On Wed, Aug 22, 2012 at 11:42 AM, Guillermo Winkler > <gwink...@inconcertcc.com> wrote: > Xu, what's your configuration? > > How many CF, how much data (size/rows/cols), how many clients operations/sec > and how much memory assigned for the heap? > > Guille > > > On Wed, Aug 22, 2012 at 12:09 AM, Xu Renjie <xrjxrjxrj...@gmail.com> wrote: > Hi, all > I have a problem about the log. I have set the CASSANDRA_HEAPDUMP_DIR in > the cassandra-env.in file to some path to store the dump file > when OutOfMemory exception happens. But after out of memory happens(I judge > it from the /var/log/messages which says "kernel: out of memory: > kill process XXXX), there is no dump file generated in that path. Is this > because kernel kill the jvm first so no OutOfMemory exception thrown or what? > I am new to linux. Thanks in advance. > > >
