Joe Elliott

I forgot about this issue! We do have both rpms and debs available for all releases: https://github.com/grafana/tempo/releases/tag/v1.5.0 Added here: https://github.com/grafana/tempo/pull/1533 Perhaps we can close this issue?

Thanks for the report! This may be a Grafana bug. > Perform Operations (Read/Write/Others) Send Traces to tempo using OTLP gRPC. After sending I can query the spans with the...

After digging some I have found that the new Parquet format is not storing links correctly, but the default format in 1.5.0 (v2) is storing them correctly. Do you know...

> Wouldn't Grafana query the frames instead of raw trace Data? I think Tempo is returning the wrong frame (without references). Grafana queries the same endpoint that you curled to...

This already exists in our enterprise tracing product: GET. We have been discussing open sourcing it, but we weren't sure if the community was interested. Thanks for the feature request!...

Reviewing the code, it looks like `tempo_ingester_live_traces{}` should be showing 20k if you are receiving that particular error message. There is some discrepancy between using the `len()` of a map...

So of the times you posted, it does look like the final one (31th @ 10:44 - 11 hits) corresponds to a spike in `ingester_live_traces_total{}`. However the other two aren't...

So it occurs to me that there is a chance, depending on your live traces flushed interval and prometheus scrape interval, to hit 20k but not actually see it on...

In Tempo objects are generally treated as immutable. I suppose there are two possible exceptions depending on what s3 considers "immutable". - When we create a block we do a...

Thanks for reporting this issue! There are a lot of factors that would result in increased memory on startup. Can you share your config? Also, in the issue you say...