ganghe / o2locktop Goto Github PK

I try running o2loktop on one of the nodes with

root@yukon:/home/rsand/o2locktop# o2locktop /mnt/stor2/data2a

o2locktop: error: can't find the mount point: /mnt/stor2/data2a, please cheack and retry

But this is where the ocfs2 filesystem is mounted.

check and guide user to config passwordless ssh for all nodes

server-out-side-of-cluster> ./o2locktop -n node1:/mnt/shared -n node2 -n node3

Ideally, the intuitive tool should check passwordless ssh for all nodes, otherwise guide user with ssh-copy-id commands. At lease, provide user the knowledge to run prepared commands, ie,

server-out-side-of-cluster> ssh-copy-id root@node1
server-out-side-of-cluster> ssh-copy-id root@node2
server-out-side-of-cluster> ssh-copy-id root@node3

I create 1,000,000 files on ocfs2 file system, and cat the locking_stat file, There is almost 500,000 item in the file. But the tool get stuck. I use o2locktop in remote mode and I check the cat option that executed by ssh, it takes about 15s. Then it takes about 2 hours to decode the locking_stat file which is unbearable.
And we should check the python sort method and the sort method in the tools to compare and get the good one.
What's more, when process too large number of locks. The CPU usage rate will up to 98%, and memory usage rate will up to 40%.

When an active file at a certain moment is no longer active at the next moment, but since the file was once very active, then even if all the files have stopped at the next moment (there is no lock on the file), the file is still Keeping a higher score, we should use a method to eliminate this situation.