Now on the ceph cluster created a piece of equipment, on a cluster node to write the piece of equipment operation, but can only be written more than 30 M per second, and they are written for a while can't write in, how to identify bottlenecks in where?

CodePudding user response:

I also like you, just start write speed can also, dozens of times, then become a few megabytes, couldn't get on, the physical machine IO delay

CodePudding user response:

reference 1st floor lzf222 response:

I also like you, just start write speed can also, dozens trillion, became a few megabytes, couldn't get on, the physical machine IO delay large

To upgrade the kernel,