Memory allocation error

wes10 · March 11, 2022, 10:18am

I am receiving this error for one of my grafana dashboard queries:
“invalid: memory allocation limit reached: limit 4000000000 bytes, allocated: 3999969472, wanted: 34176”

I would like to know if it is possible to increase the memory allocation limit?

Im using grafana cloud, influxdb cloud for the datasource, and querying with flux. Found this github issue which seems to be the same

github.com/influxdata/influxdb

InfluxDB 2.0 cloud: invalid: memory allocation limit reached: limit 10000000 bytes, allocated: 9999952, wanted: 64

opened 10:47AM - 02 Nov 20 UTC

closed 08:59PM - 03 Mar 22 UTC

paulfirstimpression

__Steps to reproduce:__ List the minimal actions needed to reproduce the behavior. 1. Use a dashboard in Grafana that has been running for weeks without problems 2. Refresh 3. Get random memory allocation limit errors __Expected behavior:__ No errors, just measurements __Actual behavior:__ I've got a Grafana dashboard with 7 graphs/tables that query InfluxDB 2.0 Cloud . Since this weekend, sometimes randomly I get _invalid: memory allocation limit reached_ errors on random queries. Refreshing the dashboard solves is, or sometimes makes other graphs give messages like these: invalid: memory allocation limit reached: limit 10000000 bytes, allocated: 9999952, wanted: 64 3 out of 7 queries take a bit longer, about 1-2 seconds, but ran without problems the last couple of weeks. Has something changed in InfluxDB 2.0 cloud over the last couple of days? Might it be that InfluxDB 2.0 can't handle the 7 queries simultaneously (dashboard refresh) anymore? __Environment info:__ * InfluxDB 2.0 cloud * Grafana 7.2.1 ( and later 7.3.1 to test any difference, but no difference)

melori.arellano · March 11, 2022, 9:56pm

welcome to the community @wes10!

Thank you for sharing the influxdb issue you found, that does look like a good match. Do you see the same message in influxdb if you run the query there? I’m suspicious that grafana is just passing the error through that it got as a response from influxdb.

wes10 · March 14, 2022, 2:50pm

Ah you are correct. I see the same message in InfluxDB. I’ll ask them what’s going on!