Running pipelines shouldn’t mean squinting at walls of raw text. With Meltano v4, structured logging and human-readable output come built in- so your team and the AI tools you rely on spends less time decoding logs and more time acting on them.
Your Logs Were Busy. Your Team Was Paying for It.
Pipeline logs exist to tell you what’s happening inside your data workflows. But in earlier versions of Meltano, that information was hard to act on. Metrics arrived embedded inside raw JSON blobs. Log lines were difficult to scan in the UI. Debugging a failed pipeline often meant cross-referencing multiple outputs just to answer a simple question: what went wrong, and when?
As data operations grow: more pipelines, more connectors, more stakeholders- that friction adds up. Teams lose time. Issues get missed. And confidence in the data platform takes a hit.
The same complexity also makes it harder for AI assistants to help. Whether you’re using an AI coding assistant or an observability platform with AI-powered analysis, clean and consistent logs produce far more useful insights than noisy, unstructured output.
The core insight: Your logs shouldn’t require interpretation. They should tell you, plainly, what happened. Meltano v4 is built around that idea.
Everything We Changed and Why It Matters
Meltano v4 brings a new default log format that is both machine-friendly for tools like Datadog and human-friendly for everyone reading them in the Meltano UI. Here is what we shipped:
- JSON structured logging enabled for all workspaces. Logs now follow a consistent schema, making them easy to index, search, and alert on- whether you are using Datadog or reading them directly.
- Readable log output reformatted in the UI. JSON log lines are translated into clean, scannable line events- not raw serialised text.
- Metrics surface inline alongside log output. Record counts and other pipeline metrics appear in context, making them easier to understand for both engineers and AI-powered log analysis tools.
- Workspace-level version control. Admins can choose between Meltano v3.9 and v4.2 per workspace from the Settings page, letting teams upgrade at their own pace.
- Python 3.10 support across all Singer SDK plugins. Connectors including tap-hubspot have been upgraded to Singer SDK v0.50.0 or higher, ensuring full structured log compatibility.
From JSON Noise to a Single Glance
Here is what the same pipeline run looks like across versions. Both are real outputs: the difference is how much work your team has to do to understand them.
V3.9- Before:
tap-meltano-cloud METRIC {"type": "counter", "metric": "record_count", "value": 42, "tags": {"stream": "accounts"}}
v4.2- Now:
tap-meltano-cloud METRIC metric_name=record_count metric_value=42
The v4 format is designed to be immediately scannable. Key-value pairs stand on their own. There is no JSON to parse, no structure to unwrap, just the information you need.
That simplicity also benefits AI assistants, which can summarize pipeline activity and identify issues more reliably without relying on custom parsing logic.
Pipeline Metrics, Right Where You’re Already Looking
Previously, metric output defaulted to stdout only, making it easy to miss in a busy log stream. In v4, metrics are emitted as structured log events alongside the rest of your pipeline output.
Live Pipeline Log- Meltano v4.2:
The bar plot in your workspace reflects record counts from these metric events in real time, giving you an at-a-glance confirmation that data moved as expected.
2026-06-19 14:03:01 tap-meltano-cloud INFO Starting extraction from stream: accounts
2026-06-19 14:03:04 tap-meltano-cloud METRIC metric_name=record_count metric_value=42
2026-06-19 14:03:04 tap-meltano-cloud METRIC metric_name=batch_size metric_value=500
2026-06-19 14:03:05 target-postgres INFO Loaded 42 records into accounts
2026-06-19 14:03:05 meltano INFO Pipeline completed successfully
One Error Event in Datadog, Not Twenty
Before this release, a single plugin failure could generate a flood of Datadog log events -one for every line the plugin emitted as it failed. That made triage painful, alert fatigue worse and cost unnecessary log ingestion.
With Singer SDK v0.50.0 and the structured logging changes in v4, plugins now emit a single, well-formed error event when something goes wrong. The event carries the full exception context- plugin name, error type, message- making it straightforward to route to the right on-call team.
Upgrade on Your Schedule, Not Ours
Not every workspace needs to move at the same time. Workspace admins can now select either Meltano v3.9 or v4.2 directly from the Advanced section of the Settings page. Run v4 in development while keeping production on v3.9- with no risk of forcing a platform-wide change on teams that aren’t ready.
Note: Meltano v4 requires Python 3.10 or higher. Check connector compatibility before upgrading if your workspace uses connectors that rely on older Python environments. All Meltano-maintained Singer SDK connectors have already been updated.
Up and Running in Minutes
Go to Settings-> Advanced, select Meltano v4.2 from the version dropdown, and save. Trigger any pipeline- logs will immediately appear in the new format. Optionally, add a Datadog API key in Workspace Settings to confirm that error events from failed plugins arrive as a single, structured entry.
See Your Pipelines Clearly. Fix Problems Faster.
Meltano v4 is a meaningful change to how your team experiences pipeline observability- from the moment a run starts to the moment it finishes and especially when something goes wrong.
Cleaner logs reduce the time between an issue occurring and your team understanding it. Structured metrics confirm data is moving correctly without a separate investigation. Because the fastest way to fix a data problem is to see it clearly in the first place.
Ready to upgrade?
Meltano v4 is available now! Enable it from your workspace settings in a few clicks and get observability that’s ready for the next generation of AI-assisted operations.