meta data for this page
  •  

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Next revision		Previous revision
vm:proxmox:ceph:performance_monitoring [2020/11/24 08:08] – created niziakvm:proxmox:ceph:performance_monitoring [2021/04/08 11:44] (current) – niziak
Line 1: Line 1:
 ====== CEPH performance monitoring ====== ====== CEPH performance monitoring ======
 +
 +===== basic info =====
 +
 +==== ceph ====
 +
 +<code bash>
 +ceph -s
 +ceph -w
 +ceph df
 +ceph osd tree
 +ceph osd df tree
 +</code>
 +
 +==== rados ====
 +
 +<code>
 +rados df
 +</code>
 +Where:
 +  * **USED COMPR**: amount of space allocated for compressed data (i.e. this includes comrpessed data plus all the allocation, replication and erasure coding overhead).
 +  * **UNDER COMPR**: amount of data passed through compression (summed over all replicas) and beneficial enough to be stored in a compressed form.
 +
 +==== RBD Rados Block Device ====
 +<code>
 +rbd ls
 +rbd du
 +</code>
 +==== perf ====
  
 <code bash> <code bash>
 ceph iostat ceph iostat
 +
 +ceph osd perf
  
 rbd perf image iostat  # and wait 30 sec rbd perf image iostat  # and wait 30 sec
 +rbd perf image iotop   # and wait 30 sec
  
 rbd du rbd du
 </code> </code>
 +
 +
 +https://ceph.io/community/new-mimic-iostat-plugin/
 +
 +==== rados benchmark ====
 +Hints from [[https://yourcmc.ru/wiki/Ceph_performance#Test_your_Ceph_cluster]]:
 +  * Don't use `rados bench`. It creates a small number of objects (1-2 for a thread) so all of them always reside in cache and improve the results far beyond they should be.
 +  * You can use `rbd bench`, but fio is better.
 +
 +<code bash>
 +ceph osd pool create test
 +rados -p test bench 10 write --no-cleanup
 +rados -p test bench 10 seq
 +rados -p test bench -t 4 10 seq
 +rados -p test bench 10 rand
 +
 +rados bench -p test 60 write -b 4M -t 16 --no-cleanup
 +rados bench -p test 60 seq -t 16
 +rados bench -p test 60 rand -t 16
 +
 +rados -p test cleanup
 +</code>
 +
 +
 +
 +==== rbd benchmark ====
 +
 +[[https://access.redhat.com/documentation/en-us/red_hat_ceph_storage/1.3/html/administration_guide/benchmarking_performance#block_device|Chapter 9. Benchmarking Performance]]
 +
 +<code bash>
 +# Image 1G:
 +rbd create test/myimage --size 1024
 +
 +#TBD ... 
 +#rbd device map test/myimage --cluster 
 +#rbd device unmap /dev/rbdX
 +
 +rbd rm test/myimage
 +</code>
 +
 +
 +