Repository navigation
Incorrect VM IOPS metrics #11421
Description
problem
VM metrics show incorrect IOPS. This may be related to the type of my primary storage. I’m using Linstor. When we perform a fio test, we get the following results:
But VM metrics shows in this time:
versions
ACS 4.19.3
Primary storage - Linstor 1.31.3
OS: Ubuntu 24.04.2
The steps to reproduce the bug
Just perform a fio test and check the metrics.
What to do about it?
It would be great if this metric could be fixed.
Activity
UAnton commentedon Aug 9, 2025
Same promlem...
DaanHoogland commentedon Aug 13, 2025
Are you also using Linstor @UAnton ?
@rp- is this a known issue?
UAnton commentedon Aug 13, 2025
@DaanHoogland hi,
No, I'm was try on NFS when deploy test cluster and get same issue
rp- commentedon Aug 14, 2025
I don't think this is related to Linstor, as I think these IOP numbers come from libvirt, as I don't recall implementing any IOP statistics.
SviridoffA commentedon Aug 14, 2025
Not sure this related, but just as additional information, my management servers has a non-En locale. I wanted to mention this because we’ve CPU metric issues related to locale in the past.
DaanHoogland commentedon Aug 14, 2025
ok, so this is not storage type related. What does the graph below this one say, @SviridoffA ? the one that has reads and writes seperated in different values. I am wondering if this is a stats collector issue, a UI issue or a historic data/accumulation issue.
SviridoffA commentedon Aug 14, 2025
@DaanHoogland
Results, including the Read & Write graph, are below. Looks like incorrect too.
DaanHoogland commentedon Aug 14, 2025
I’ve seen people show behaviour like this before.
I think this one will be hard to pinpoint.
EduFrazao commentedon Sep 4, 2026
Samething here with Cloudstack 4.22.1.0 and Ceph 20.2.3
Data reading and writing metrics seems fine!