-
In my current setup, there are some steps between ingesting data and producing the final data product table.
For example, I don't treat intermediate tables as data products. Instead, I created data…
-
Ingestion Client and Server: Version 2
Work Package Overview
This work package outlines the changes required to make a more stable, less error prone version of the client that will make it more modul…
-
There is a need to ingest OTLP metadata. After the upstream Prometheus pr gets merged, add this feature to the Cortex.
Upstream Prometheus pr: https://github.com/prometheus/prometheus/pull/13439
-
## Description
NO2 (#89) and Geoglam (#167, #173) datasets requires monthly ingestion as new assets are created. This is currently a manual process however should be automated. `veda-data-airflow` ha…
-
## Process for Ensuring Package Ingestibility
1. **Request a score for a given package**:
- Initiate a request to evaluate the package by calling our scoring endpoint.
2. **Call the endpoint…
-
### Which OpenObserve functionalities are relevant/related to the feature request?
other
### Description
In the current UI, the label “Ingestion” is being used to represent data sources. To make th…
-
## Description
When in my playbook I have a "manipulate knowledge" to replace the Author, nothing happens. By this I mean that nothing is created / updated in my platform while in the traces of the p…
-
When try to load batch ingestion job with hadoop job is not getting started because jetty classes loading from Pinot batch ingestion shaded jar could not able to find the webapp files which leads to …
-
| Start Date | ETA | Project Lead | Actual Ship Date |
| ---------- | ---------- | ------------ | ---------------- |
| 2024-03-14 | | @stacimc | TBD |
## Description
We…
-
I'm looking into the AIND format option for inputting data into the pipeline. The readme states
> aind: data ingestion used at AIND. The input folder must contain an ecephys subfolder which in turn…