--disable-metrics case

[romantico88]

Hello, i saw that there is posibillity to disable some metrics.
I want to disable f.ex openstack_nova_server_status, openstack_cinder_status. When i'm using --disable-metric=openstack_server_status, it's not working.

Which names of metrics should i use? those from openstack ceilometer/metrics services or those from exporter? how can i see it?

[alexeymyltsev]

Hello,
As mention in description it should be as service name "-" metric name:
--disable-metric can be specified in the format: service-metric (i.e: cinder-snapshots)
It mean you should have
-d nova-server_status -d cinder-volume_status

[romantico88]

i've also try this, exporter is launched, but after making curl to metrics after 11-12 seconds have error:
(it's happens with disabling nova-server_status)

: panic: runtime error: invalid memory address or nil pointer dereference

lip 23 11:26:06 openstack-exporter[8605]: panic: runtime error: invalid memory address or nil pointer dereference
lip 23 11:26:06 openstack-exporter[8605]: [signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x8f11f3]
lip 23 11:26:06 openstack-exporter[8605]: goroutine 28 [running]:
lip 23 11:26:06openstack-exporter[8605]: github.com/openstack-exporter/openstack-exporter/exporters.ListAllServers(0xc000070ba0, 0xc000070720, 0xf1eb40, 0x2)
lip 23 11:26:06openstack-exporter[8605]: /app/exporters/nova.go:289 +0x4f3
lip 23 11:26:06openstack-exporter[8605]: github.com/openstack-exporter/openstack-exporter/exporters.(*BaseOpenStackExporter).RunCollection(0xc000070ba0, 0xc00042bbd0, 0xa60f25, 0x9, 0xc000070720, 0x0, 0x0)
lip 23 11:26:06openstack-exporter[8605]: /app/exporters/exporter.go:98 +0x20c
lip 23 11:26:06openstack-exporter[8605]: github.com/openstack-exporter/openstack-exporter/exporters.(*BaseOpenStackExporter).Collect(0xc000070ba0, 0xc000070720)
lip 23 11:26:06 openstack-exporter[8605]: /app/exporters/exporter.go:121 +0x126
lip 23 11:26:06 openstack-exporter[8605]: github.com/prometheus/client_golang/prometheus.(*Registry).Gather.func1()
lip 23 11:26:06  openstack-exporter[8605]: /go/pkg/mod/github.com/prometheus/[email protected]/prometheus/registry.go:443 +0x19d
lip 23 11:26:06  openstack-exporter[8605]: created by github.com/prometheus/client_golang/prometheus.(*Registry).Gather
lip 23 11:26:06 openstack-exporter[8605]: /go/pkg/mod/github.com/prometheus/[email protected]/prometheus/registry.go:454 +0x57d

using cinder-volume_status

its no error, but openstack_cinder_volume_status is still presented on metrics.

used newest version - downloaded amd64 tar.gz , centos 7 / centos 8.

[romantico88]

also i used a default cinder-snapshots like i example and metric still presented in exports.

i running exporter like below:

/usr/bin/openstack-exporter --os-client-config /etc/openstack-exporter/clouds.yaml --web.listen-address=":9170" mycloud --log.level="debug" -d cinder-snapshots

time curl http:https://localhost:9170/metrics 2>&1 | grep snap 
# TYPE openstack_cinder_snapshots gauge
openstack_cinder_snapshots 23

and my question is what i doing wrong?:)

[alexeymyltsev]

i've also try this, exporter is launched, but after making curl to metrics after 11-12 seconds have error:
(it's happens with disabling nova-server_status)

: panic: runtime error: invalid memory address or nil pointer dereference

lip 23 11:26:06 openstack-exporter[8605]: panic: runtime error: invalid memory address or nil pointer dereference
lip 23 11:26:06 openstack-exporter[8605]: [signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x8f11f3]
lip 23 11:26:06 openstack-exporter[8605]: goroutine 28 [running]:
lip 23 11:26:06openstack-exporter[8605]: github.com/openstack-exporter/openstack-exporter/exporters.ListAllServers(0xc000070ba0, 0xc000070720, 0xf1eb40, 0x2)
lip 23 11:26:06openstack-exporter[8605]: /app/exporters/nova.go:289 +0x4f3
lip 23 11:26:06openstack-exporter[8605]: github.com/openstack-exporter/openstack-exporter/exporters.(*BaseOpenStackExporter).RunCollection(0xc000070ba0, 0xc00042bbd0, 0xa60f25, 0x9, 0xc000070720, 0x0, 0x0)
lip 23 11:26:06openstack-exporter[8605]: /app/exporters/exporter.go:98 +0x20c
lip 23 11:26:06openstack-exporter[8605]: github.com/openstack-exporter/openstack-exporter/exporters.(*BaseOpenStackExporter).Collect(0xc000070ba0, 0xc000070720)
lip 23 11:26:06 openstack-exporter[8605]: /app/exporters/exporter.go:121 +0x126
lip 23 11:26:06 openstack-exporter[8605]: github.com/prometheus/client_golang/prometheus.(*Registry).Gather.func1()
lip 23 11:26:06  openstack-exporter[8605]: /go/pkg/mod/github.com/prometheus/[email protected]/prometheus/registry.go:443 +0x19d
lip 23 11:26:06  openstack-exporter[8605]: created by github.com/prometheus/client_golang/prometheus.(*Registry).Gather
lip 23 11:26:06 openstack-exporter[8605]: /go/pkg/mod/github.com/prometheus/[email protected]/prometheus/registry.go:454 +0x57d

using cinder-volume_status

its no error, but openstack_cinder_volume_status is still presented on metrics.

used newest version - downloaded amd64 tar.gz , centos 7 / centos 8.

This panic happened because some of metrics collected in one function. It mean function executed , but if some metic has been desalted then it can not be written and panic happened.

We should work on it. We should create new issue about it.

[alexeymyltsev]

also i used a default cinder-snapshots like i example and metric still presented in exports.

i running exporter like below:

/usr/bin/openstack-exporter --os-client-config /etc/openstack-exporter/clouds.yaml --web.listen-address=":9170" mycloud --log.level="debug" -d cinder-snapshots

time curl http:https://localhost:9170/metrics 2>&1 | grep snap 
# TYPE openstack_cinder_snapshots gauge
openstack_cinder_snapshots 23

and my question is what i doing wrong?:)

I have tested several times mostly same string
go run *.go --os-client-config=./clouds.yaml --disable-cinder-agent-uuid default --log.level="debug" -d cinder-snapshots
and it works without any problem

time curl http:https://localhost:9180/metrics 2>&1 | grep snap
curl http:https://localhost:9180/metrics 2>&1  0,00s user 0,01s system 0% cpu 18,888 total
grep --color=auto --exclude-dir={.bzr,CVS,.git,.hg,.svn} snap  0,01s user 0,00s system 0% cpu 18,888 total

[jvleminc]

I am facing the same issue, related to another issue I created: #208

Containers starts up well, but when calling /metrics the goroutine error is thrown:

panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x8cbc66]

goroutine 42 [running]:
github.com/openstack-exporter/openstack-exporter/exporters.ListZonesAndRecordsets(0xc0004b40f0, 0xc0004ff200, 0xedc1e0, 0x2)
	/app/exporters/designate.go:95 +0xc36
github.com/openstack-exporter/openstack-exporter/exporters.(*BaseOpenStackExporter).RunCollection(0xc0004b40f0, 0xc0004be080, 0xa399e7, 0x5, 0xc0004ff200, 0x0, 0x1)
	/app/exporters/exporter.go:106 +0x1eb
github.com/openstack-exporter/openstack-exporter/exporters.(*BaseOpenStackExporter).Collect(0xc0004b40f0, 0xc0004ff200)
	/app/exporters/exporter.go:127 +0x101
github.com/prometheus/client_golang/prometheus.(*Registry).Gather.func1()
	/go/pkg/mod/github.com/prometheus/[email protected]/prometheus/registry.go:443 +0x19d
created by github.com/prometheus/client_golang/prometheus.(*Registry).Gather
	/go/pkg/mod/github.com/prometheus/[email protected]/prometheus/registry.go:535 +0xe12

Update: I see #200 exists for this issue; will comment there.

[LaurentDumont]

I see the same issue with disabling --disable-metric nova-server_status. We'd like to remove the data because it's obviously a really expensive metric with bigger clouds.

[frmorais]

Is this fixed? I'm having the same issue when using --disable-slow-metrics

[jvleminc]

@frmorais I closed #208 after upgrading worked.

[frmorais]

@jvleminc In my case I'm still getting the 'panic: runtime error' and I'm using latest docker image available.

EDIT:
It seems to me that image_bytes is considered a slow metric. So, when using --disable-slow-metrics it will give an error, because cannot collect the other glance metrics :

panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x85261a]
goroutine 131 [running]:
github.com/openstack-exporter/openstack-exporter/exporters.ListImages(0xc0002b60c0, 0x2f?)
/exporters/glance.go:54 +0x39a
github.com/openstack-exporter/openstack-exporter/exporters.(*BaseOpenStackExporter).RunCollection(0xc0002b60c0, 0xc00039db50, {0x97c14e, 0x6}, 0x0?)
/exporters/exporter.go:99 +0x182
github.com/openstack-exporter/openstack-exporter/exporters.(*BaseOpenStackExporter).Collect(0xc0002b60c0, 0xc00059bf60?)
/exporters/exporter.go:122 +0x105
github.com/prometheus/client_golang/prometheus.(*Registry).Gather.func1()
/go/pkg/mod/github.com/prometheus/[email protected]/prometheus/registry.go:443 +0xfb
created by github.com/prometheus/client_golang/prometheus.(*Registry).Gather
/go/pkg/mod/github.com/prometheus/[email protected]/prometheus/registry.go:535 +0xb0b

Disabling completely image service with '--disable-service.image' fix this, but no glance metrics will be available of course.
Tested this with an older branch (v1.4.0) and this issue was not there.