Hi,
As agreed i have tried to come up with the metrics we need in order to get AFM in the grafana dashboard.
Im not sure which to choose excactly, maybe you have some information about that?
GPFSAFM, GPFSAFMFS, GPFSAFMFSET?
Which is best to use for monitoring AFM in grafana?
We usually use the command "mmafmctl getstate" to see the state, que and filesets used for AFM. sometime we use mmfsadm dump afm to see more detailed information as it also displays split write, current files in que/inflight (per node) and much more.
My best guess would be these are required as a minimum, but not 100% sure:
gpfs_afm_queued
gpfs_afm_inflight
gpfs_afm_complete
gpfs_afm_errors
i think the above is a good pointer towards checking the health of AFM on the different filesets?
Here is my initial idea of metrics to use for basic performance monitoring:
gpfs_afm_bytes_read
gpfs_afm_bytes_written
gpfs_afm_ops_sync
gpfs_afm_ops_sent
gpfs_afm_bytes_pending
gpfs_afm_avg_time
gpfs_afm_tot_write_time
gpfs_afm_conn_broken
gpfs_afm_used_q_memory
gpfs_afm_num_queued_msgs
if you have some better ideas im open for it.. we just need to make sure that we can see the current state of AFM in grafana as well as monitor the que, performance etc..
/Andi Christiansen