Dashboard times out when more than about 24 hours of data is queried

[yosiasz]

Are you using same time range

[lifeboy]

Yes. 30 days.

[lifeboy]

So does that mean that I have to gives up. The query runs fine for 30 days in InflxuDB, but times out in Grafana.

Can one open a bug submission for Grafana?

[yosiasz]

@lifeboy

120 seconds

Before you do that…

120 seconds is still an issue for a 30 day span in any database. You should be getting subseconds speed if not few seconds only with some finetuning.

Next step is vetting your influxdb schema design and the use of the contains filter

The rule with influxdb is as follows:
measurement and tags are indexed whereas fields are not.
So you may want to configure object as a tag (if not already) since you are filtering on it

InfluxDB schema design and data layout | InfluxDB OSS v1 Documentation

The other issues you will need to tackle is the use of contains which is notorious for performance issues in any query language. Other alternatives are found in this post

Grafana + InfluxDB Flux - query for displaying multi-select variable inputs

We’ve set up a Grafana cloud + InfluxDB 2.0 (Flux language) cloud instance. As part of this, we’ve created a custom variable list with “device IDs”, called devices . In a panel, we wish to display parameter data where the user can select one or more device IDs from the devices list to have them displayed in the panel. This works fine for single device ID selection, but not for multiple devices. How should the query be modified to display data from a variable number of devices based on a mul…

So once you have done your due diligence with best practices on the influx schema side of things and using something other than contains function, you can then confidently create a bug for increasing timeouts

[lifeboy]

@mephisto, you created this dashboards for Proxmox! Where are you when I need you?

[yosiasz]

@lifeboy

I really don’t need to understand the full dashboard if I’m just a user of it, do I?

did you setup the influxdb yourself?

[lifeboy]

Yes, I did. I following the instructions to set it up for Proxmox. I’m busy learning basics of the query language and have fixed on of the graphs already by making the changes you linked me to.

New query using regex:

filter(fn: (r) => r[“nodename”] =~ /${server:regex}/)

which was before:

filter(fn: (r) => contains(value: r[“nodename”], set: ${server:json}))

Can’t wait for @mephisto forever now, can I?

[yosiasz]

nope. is nodename a field or a tag?

create a bucket + token in influxdb2
enable metric export in Proxmox VE. <--- this section is a concern

how much faster is your query now with the regex change

[lifeboy]

I see there is a new version of the Dashboard at Proxmox Cluster [Flux] | Grafana Labs which has dropped the “contains” from the queries.

It has a downside in that one cannot select stats for the whole cluster, but only for one host at a time, but at least I can get stats for 30 days.

So I’m going to close the thread now. My objective is not to learn Flux at this stage, but to simply use the dashboard, so the pressing problem has been resolved.