Skip to main content

Pipeline

The Pipeline screen lets you check how well SingleStore's data ingestion pipelines are running. When a pipeline falls behind or fails, data does not arrive on time. This screen brings together throughput, latency, and errors so that you can quickly pinpoint where the problem is.

What is a pipeline

A SingleStore pipeline automatically pulls data from external sources (files, message queues, and so on) and loads it into tables. In simple terms, it is like a conveyor belt that continuously moves data. If the belt slows down or stops, data piles up, so it is important to watch the processing speed and errors.

Basic screen guide

Select the cluster to query at the top of the screen. The screen consists of a status summary, KPI cards, trend charts, a pipeline details table, and recent errors.

KPI cards

The cards show the key metrics of the overall pipeline.

CardDescription
PIPELINESTotal number of pipelines
FAILING PIPELINESNumber of pipelines that are currently failing
BATCH SUCCESS RATEBatch success rate
FAILED BATCHESNumber of failed batches
BATCHESNumber of processed batches (batches per interval)

Each card also shows a secondary unit such as rows/s (avg), batches per interval, or milliseconds (avg).

Trend charts

Two time-series charts show the processing status of the pipelines.

ChartDescription
RPS by PipelineRows processed per second for each pipeline (rows per second)
Batch Success / FailureTrend of batch successes and failures

Clicking a specific pipeline in a chart filters both the chart and the table below by that pipeline.

Pipeline Details table

The table shows the processing status of each pipeline. You can search by pipeline, source, or table name in the search box at the top.

ColumnDescription
PipelinePipeline name
SourceData source
Target TableTarget table for ingestion
TypeSource type
KindPipeline kind
RPS (avg)Rows processed per second (average)
Rows TotalTotal number of rows ingested
Batches OKNumber of successful batches
FailedNumber of failed batches
QueuedNumber of queued batches (backlog)

Clicking a pipeline row in the table opens the batch-level details drawer. In the drawer, you can review the batch metric charts and recent errors for the selected pipeline.

Recent Pipeline Errors

At the bottom of the screen, you can check the pipeline errors that occurred recently.

ColumnDescription
Error TimeTime the error occurred
PipelinePipeline where the error occurred
BatchBatch identifier
CodeError code
MessageError message
HostHost where it occurred
InstanceInstance where it occurred
  • Opening the batch drawer automatically queries recent errors.
  • The error list is shown across multiple pages.
  • You can search by pipeline, type, code, or message in the search box.

Query time

You can check the last collection time at the top right of the screen.