What’s new in Prometheus 2.28?

Published 08.07.2021

Categories Changelog

Prometheus 2.28 is out. If you don’t know, Prometheus is an excellent open-source system monitoring and alerting toolkit. Let’s have a look at those features and have a look at the changelog.

Displaying Trace Examplers in the Graphic Interface

From the previous versions of 2.26 and 2.27, we can see the new support that Prometheus received for storing and receiving trace exemplars and sending exemplar data over Prometheus’s remote_write interface.

In this version, we see that the integration of exemplars with metrics is taking a step further with added support for displaying tracing exemplars in Prometheus’s graphing interface. From now on, whenever a PromQL query accesses series with exemplar data attached to them, we can now show these exemplars along with the returned PromQL query data by enabling a per graph “Show Exemplars” setting.

prometheus 2.28

Per Graph Show Exemplars (Source: Promlabs)

A latency-related graph allows us to quickly find trace exemplars for requests with high latency to investigate those requests in a tracing system further. Suppose we want to copy part of the exemplar metadata (such as the trace ID) to correlate it with another system in production. In that case, we can click on any exemplar to show its metadata more persistently under the graph.

An easy way for us to see this is to get a demo running where we can try out the exemplar display is by running Grafana’s “The New Stack” demo app:

# Clone the "The New Stack" demo app.
git clone git@github.com:grafana/tns

# Check out a predictable version of the repo.
git checkout ec5673d

# Use Prometheus 2.28 instead of 2.26.
cd production/docker-compose
sed -i 's/prometheus:v2.26/prometheus:v2.28/' docker-compose.yml

# Run components necessary to expose and store metrics and trace exemplars.
docker-compose up tempo db app loadgen prometheus

After this, we can head to  http://localhost:9090/ and run the query:

histogram_quantile(0.99, sum by(le, method) (rate(tns_request_duration_seconds_bucket[1m])))

We can click the “Show Exemplars” button to show trace exemplars.

Generic HTTP-Based Service Discovery

We have seen earlier that Prometheus has supported generic service discovery integrations for a long time via its file_sd discovery mechanism, which watches and reads a set of target files from disk. The only downside of the file_sd approach is that it requires sharing a filesystem between Prometheus and a sidecar process that generates the target files, which can be problematic in some environments. Users have asked for a network-based alternative for a long time.

A new HTTP-based generic discovery mechanism is now in Prometheus v2.28. This mechanism works almost like file_sd but loads target information from a specified URL rather than from a local file:

- job_name: http-sd
    - url: 'https://infra-db/prometheus-sd'

It also only supports JSON as the targets format, whereas file_sd supports both JSON and YAML.

This new HTTP-based discovery mechanism not only encourages people to build remote sidecar processes that can pull target information from various sources of truth. But developers will add native Prometheus HTTP discovery support directly into their infrastructure and service databases, getting rid of the need to run a separate process.

Defaulting to the New Expression Editor

During Prometheus 2.26, we saw the introduction of a shiny new PromQL editor with advanced auto-completion, inline linting, and syntax, highlighting that you had to explicitly enable it as an experimental feature. Since this new editor is so much more usable than the previous bare text input field, and since no major complaints came about it, Prometheus 2.28 now defaults to this new editor:

New Editor (Source: Promlabs)

This default is stored in the browser’s local storage and is only set to the new default value if no previously stored value is present for it yet. Thus if we have used Prometheus 2.26 or 2.27 in the meantime and now access 2.28 using the same URL, we may still have to enable the new editor explicitly.

Other Features and Enhancements

Promtool: We will now see an allowance of silencing output when importing or backfilling data.

Consul SD: It will now support reading tokens from the file.

Rules: Added a new .ExternalURL alert field templating variable containing the external URL of the Prometheus server.

Scrape: Added experimental body_size_limit scrape configuration setting to limit the allowed response body size for target scrapes.

Kubernetes SD: Added ingress class name label for ingress discovery.

UI: Showing of a startup screen with a progress bar when the TSDB is not ready yet.

SD: Addition of a target creation failure counter like prometheus_target_sync_failed_total and improvement of target creation failure handling.

TSDB: Improving validation of exemplar label set length.

TSDB: Added a prometheus_tsdb_clean_start metric indicating whether a TSDB lockfile from a previous run still existed upon startup.

Bug Fixes

UI: In the experimental PromQL editor, fixing autocompletion and parsing for unique float values and improving series metadata fetching.

TSDB: When merging chunks, split resulting chunks would contain more than the maximum of 120 samples.

SD: Fixed the computation of the prometheus_sd_discovered_targets metric when using multiple service discoveries.


A lot of new things came with Prometheus v2.28. You can try out this new version by clicking here.

Read more posts on CNCF here:

Join 100+ cloud native enthusiasts

and stay in the loop on modern software development.

Sign up to receive exclusive content around cloud native software development right into your inbox.

We don’t spam! Read our privacy policy for more info.

More stories from our blog

What’s new in Harbor v2.3.0?

What’s new in Harbor v2.3.0?

Harbor v2.3.0 is here, and with all recent versions, it has brought several new features and has undergone some changes. Let's roll! New Features Declarative Configuration In the latest release, we can see the enhancement of the current configuration by adding...

What’s new in Kubernetes v1.21.2?

What’s new in Kubernetes v1.21.2?

It's June, and Kubernetes has released a new update with version 1.21.2. We will have a look in brief at the changes that came along with this update. We will also have a look at the bugs that Kubernetes removed ahead with the few things added. Let's roll. Changes...

Chaos Engineering: Not so Chaotic

Chaos Engineering: Not so Chaotic

It feels very complex when we talk a lot about cloud computing and developer operations. Furthermore, certain things look complicated, but they are not so if we easily understand those concepts. Today, we will discuss such a thing that sounds complex but is simple and...

On Charming Engineering Culture: My Notes

On Charming Engineering Culture: My Notes

Engineering teams are at the core of any modern organisation. They break/make an organisation, and empowering them is critical to any modern companies’ success. A motivated engineer brings more value than a ‘whatever’ engineer. Its high time managers and leaders focus...

Observability: Your Eyes in Cloud

Observability: Your Eyes in Cloud

Observability is all around the cloud. You might come across the term while exploring the vast stretches of documentations or blog posts, maybe videos or streams too. Well, from far you might have seen that this is a very broad term, and it’s expected. The topic is...

Cloud Firewalls Simplified: Beginners  Edition

Cloud Firewalls Simplified: Beginners Edition

Cloud technology is everywhere. From your photos to big corporations carrying out their day to day operations. But have you ever thought about the security needed to protect this vast pile of data? Security from external attacks by threat detection and elimination is...

Object and Block Storage: How They Differ?

Object and Block Storage: How They Differ?

The difference between block and file storage makes heads spin due to the complexity of definitions and technical jargon across the internet. Even a technical person sometimes forgets the business value and makes decision fatigue their best friend when trying to...

Helm: Why DevOps Engineers Love it?

Helm: Why DevOps Engineers Love it?

Kubernetes doesn’t have reproducibility built-in. At least, that’s what we hear most people complain as a cloud native consultation firm serving both startups and enterprises. I have been using Kubernetes for a while now, and it stands up to the mark of being a gold...

Interested in what we do? Looking for help? Wanna talk about software strategy?