Do not simply disable ballooning, since that'll force the hypervisor to start swapping out guest memory, completely killing performance. Note that ballooning is a symptom of a much bigger problem: lack of resources on the host. This can have side effects: if you have disabled swap, the guest OS will simply refuse any requests for more memory than is available. That fits rather perfectly with your 11 GB of 16 GB used.īallooning will appear as used memory in the guest - and since it's performed by a kernel-mode driver, it won't appear as a process. In fact, the default max balloon amount is supposedly 65% of the guest's physical memory. If you are allocating more memory than the host has (overcommitment), and the host is running out of memory, it'll start ballooning - basically, it'll start taking up space in the guests (forcing the guests to start swapping) so the hypervisor has more memory to work with. You should check how much memory you are allocating to each guest. I see that you are running inside VMware: Hypervisor vendor: VMware Am I missing something here?įollowing is the CPU detail: Architecture: x86_64 I am not getting which processes are eating up the memory. This is the output of the top command with PIDs sorted by RAM usage: total used free shared buffers cachedĭropping caches by echo 1 > /proc/sys/vm/drop_caches is not helping. Now, when I check the memory usage, 11GB out of 16GB is used. The process has crashed after an OutOfMemory error. I am running a cassandra instance on a server with 16GB RAM.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |