As a project manager for a small Rails agency, I find it challenging to keep track of every client’s application. Is the site live? Is it stable? Do we have a silent issue that frequently rears its head?

AppSignal makes things like anomaly detection, uptime monitoring, and issue resolution easy, even for a non-technical project manager!

Keeping An Eye on Things

We want our clients to think we’re everywhere all at once. We know the site is up. We know the code is behaving. We know the servers are happy. But, as an agency grows, it becomes increasingly difficult to keep tabs on everything. AppSignal helps by offering great out-of-the-box solutions that showcase key metrics in nice-looking, digestible graphs on premade dashboards.

As a project manager, I have some understanding of how an application works. But that doesn’t mean I know exactly what ActiveRecord does in Ruby on Rails! It’s complicated. So, how do I keep an eye on fifteen different applications at once? Don’t tell our clients this, but: I don’t!

AppSignal’s alerts and anomaly detection give me confidence that things are being monitored and can be appropriately escalated as needed. If I need to check on something, I can understand what I am looking at in the dashboard.

Here are a few helpful checks that I do at the beginning of my day to make sure that everything is working as expected:

• Check an entire organization's throughput, mean response time, and error rate percentage on their dashboard. These are easy indicators to check. If the numbers look wild and out of control, things have probably gone wrong!

• In the application view, check the latest opened errors and latest opened performance measurements for critical systems that are not set up with alerts.

• In the ActionMailer dashboard, I check the last 24 hours for any drastic spikes or complete outages. These are often tied to service interruptions or server issues for our clients.

These quick checks on AppSignal’s default dashboards help me rapidly escalate any issues to the development team, keep in touch with how a client’s application is behaving, and raise any long-running issues in our weekly meetings.

Getting Everything In One Place

One of the best things about using AppSignal is that all of our monitoring is in one place. At one point or another, Planet Argon was using BugSnag for error tracking, New Relic for performance and server monitoring, Pingdom for uptime monitoring, and manually pulling logs from the servers if there was an issue. Now, all of these features are available in AppSignal!

Selfishly, this saves me a lot of time and effort. Whenever a new client joins or a new developer is onboarded to a project, I only have to set them up in AppSignal.

Having these metrics in the same place also allows you to drill down into issues with several different entry points. If you have an outage, you can go into AppSignal and see a few things very quickly:

• Are all of the applications down, or just one? - in uptime monitoring
• Has a deployment failed or caused a considerable amount of errors?
• Is the application overloaded? See memory usage
• Can I find anything in the logs?

Then, depending on the issue, a developer can drill down deeper into the data. Let’s say there is an issue with our latest deployment. The deployment “succeeded” when processed by the server, but the application immediately crashed, and the site is now down. AppSignal allows you to filter error messages and performance issues by deployment. Then you can look for big outliers like a spike in the throughput or errors for a specific function. This takes only seconds.

A Real Example in AppSignal

The image below shows a Template Error on the show page for a product on our client’s website. This error had a cluster of spikes in late September and then was gone for a good chunk of October before recurring. This tells me that a fix may have been applied to the original issue that was modified or removed during a recent deployment. At the bottom of the graph there are “Markers” which, in this case, indicate deployments. Making this base-level assumption helps me flag a scenario to our developers and give them some data to jump off from.

Building Our Clients' Confidence with AppSignal's Data

At Planet Argon, we work with clients to maintain and improve their existing applications across multiple platforms. In practical terms, this means dedicated hours are spent tackling technical debt, making small feature improvements, and patching applications primarily written by other developers. One challenge is that you might not have all the technical details or historical knowledge you need to understand what’s happening when a new issue arises.

As a project manager, I ensure that clients have a clear picture of what an issue means for their business's larger infrastructure. Sometimes this can get technically complicated, especially when diving into systems like Cron Jobs and services like Sidekiq Pro.

What I like about AppSignal, as a project manager trying to manage client expectations, is that I can examine data on many different levels. Some clients are not technical, but they can understand when a service is humming along and then suddenly cuts out! Other clients need more detail. They may be interested in the specific service that has suddenly dropped out, or they might know exactly what ActiveRecord does in Ruby on Rails, for example.

Often, the narrative of an issue is the most critical part of delivering bad news. With features like the anomaly detection triggers in AppSignal, I can set up Slack notifications to alert the team whenever a specific issue recurs, attach it to deployment markers, and keep an eye on recurring issues in the logbook. This level of detail, alerts, and tracking helps us build trust with our clients and keeps them informed of issues we track in their applications.

A Real Scenario in AppSignal

An e-commerce client of ours had an issue where their email service would crash silently and then be down for a dozen hours before anyone noticed. This service was responsible for order confirmation emails, customer support requests, and password reset emails. Upon further investigation, we found that the email service was part of an elastic container service in AWS containing services that helped update thousands of records on their site including products, users, and other special conditions. So, if the product update service failed, so did the email service.

AppSignal allowed us to do a few things immediately:

1. Provide the client with specific details and a timeline of when the service interruption started.
2. Set up a trigger in anomaly detection to alert Slack if the same issue occurred.
3. Spin up a page for downtime monitoring to let the client check on services themselves at any point.

Investigating this issue in AppSignal was simple: we could look at the Throughput metric, find the productUpdater function, and see that it was looping requests for each database row every hour. This caused a memory overload in the container mentioned earlier and crashed it! The fix was more complicated, but that’s a story for a different day.

Bringing the Client Into the Ecosystem

There is a strong tendency in the development agency world to segregate clients from development tools as much as possible. This stems from a simple logic: many things don’t make sense if you don’t know what you’re looking at! Many errors look scary when you don’t know how to read them.

Here are some tips to help non-technical clients read error monitoring graphs like the ones provided by AppSignal:

• When there is an issue, show the client when it started to occur on the errors graph.
• Explore an Issue Sample with a client. Discuss whether the error is happening on multiple pages or only a single one. Show them the URL in the Sample view.
• Use the Issue Trend graph on a Sample to identify how often the error occurs.
• Then work with the client and lead developer on the project to attach severity levels to issues in AppSignal.
• Close issues together when a problem is resolved.

This process will not work for every client, but by previewing some error monitoring features to them, you will build their confidence that you are on top of what’s happening, can pinpoint issues, and find a path to the solution.

Happy debugging!

P.S. Contact us to learn more about AppSignal or sign up for a free trial and see how AppSignal can help you manage applications seamlessly.