-
Notifications
You must be signed in to change notification settings - Fork 215
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
OPC Publisher 2.9.4's Prometheus metrics partially missing probably due to size limit #2197
Comments
After have specified "publisherId" and "siteId" with a short string, more metrics are available:
|
Likely due to the default of MaxMetricStreams in OTEL library being 1000. You could try setting
in the Configuration.cs file of the module. If that works, please let me know. Otherwise, I have added to 2.9.5 release. |
I think we should fix the issue by removing low-value and non-documented metrics. For me increasing the size limit by 300% is not reliable: why 3000 is enough? Why not 5000 or 10000? I attached below another sample we captured in production. Half of the metrics response is filled with We care more abount "iiot_" metrics which are unfortunately almost missing We are paying Azure for transfering metrics over IoT Hub and storaging them in Log analytics workspace. We should have an option to disable the metrics which we don't need
|
If you are using |
I would really appreciate if this regression can be fixed instead of being postponed again and again to later version. We have deployed in production some Azure Monitor alert rules and dashboard relying on Prometheus metrics which are now missing since v2.9.4. This bug makes our operation team difficult to monitor the system. |
In 2.9.9 we are adding a--ari command line option to enable all automatic .net runtime instrumentation. By default it will be off. This will reduce the metrics significantly and align with previous behavior. We will also disable "total" suffix, which can be enabled by --ats command line option. Furthermore response caching will be disabled, cannot be enabled/configured. |
Describe the bug
It appears that OPC Publisher v2.9.4's prometheus endpoint (http://<opc_publisher>/metrics) does not always expose all metrics documented in this article.
It seems that the metrics are somehow truncated when the total data size exceeds certain threshold.
To Reproduce
Please find below a sample of raw metrics content.
Many iiot metrics (e.g.,
iiot_edge_publisher_messages_total
) are missing. Instead the content is filled with a lot of non-documented metrics such asdns_lookup_*
,process_runtime_dotnet_*
,kestrel_*
,http_server_request_*
, etc.When comes to iiot metrics, there are a lot of metadata filled with long but low-value default values such as
siteId="<<UnknownSite>>",writerGroupId="<<UnknownWriterGroup>>"
.In order to reduce the content size, we have configured the "publisherId" and "siteId" with shorter string, and the metrics are back afterwards.
Expected behavior
Prometheus endpoint always return all documented metrics
Screenshots
N/A
Desktop (please complete the following information):
Additional context
OPC Publisher version: 2.9.4
The text was updated successfully, but these errors were encountered: