If you have multiple processes sharing a single disk, then you can use the locality API to tag each process with an identifier for its disk and then read them back out with:
cluster.processes.<process_id>.locality.<identifier_name>
In this case, you would only count the total and free bytes once per
disk. For available bytes, you would add free bytes once per disk and
(available-free) for each process.
Hi, I’m getting confused by this part of the docs (Monitored Metrics — FoundationDB 7.1).
Can you explain how to tag each process identifier for its disk?
I have tried to define a variable locality_diskid=sdbm to identify the disk:
[fdbserver.4500]
datadir = /fdb-sdbm/data/$ID
locality_diskid = sdbm
locality_machineid = machine_129
locality_zoneid = machine_129
class = storage
Is that ok? But I’m still confused about the computation of the size.
So what should I do for the next step?