Using Mezmo To Troubleshoot In Production

4 MIN READ
MIN READ
TABLE OF CONTENTS
    4 MIN READ
    MIN READ

    In 1946, a moth found its way to a relay of the Mark II computer in the Computation Laboratory where Grace Hopper was employed. Since that time, software engineers and operations specialists have been plagued by “bugs.” In the age of DevOps, we can catch many bugs before they escape into a production environment. Still, occasionally they do, and they can spawn all kinds of unexpected problems when they do.

    In addition to software bugs, today’s modern systems encounter other problems as well. And these problems collectively manifest as issues in production. Or, as some might refer to them, the dreaded incident!

    When an incident occurs, it’s not always readily apparent why it happened or what caused it in the first place. This article will explore some of the different situations that can result in a production incident, and we’ll investigate how you can uncover these situations using Mezmo, formerly known as LogDNA. Troubleshooting a production incident shares many similarities with how a medical professional might approach a patient’s diagnosis. We look at the symptoms, we run tests, and we reach a diagnosis by drawing on our experience, the experience of others, and sometimes even relying on the process of elimination.

    Detection and Recovery

    Production incidents have two distinct phases: detection and recovery. Many teams track Mean Time To Detection (MTTD) and Mean Time To Recovery/Resolution (MTTR) metrics. MTTD is a measure of how long it takes for a problem or incident to be identified and acknowledged by the responsible team. MTTR then measures the amount of time before the team can identify and rectify the underlying cause.

    This article is not about tracking metrics; however, it is essential to measure and track their effect on our results when we’re investigating and investing in new ideas and tools. Metrics such as MTTD and MTTR provide visibility into how effectively the organization manages the incident process and how valuable different approaches and toolsets are within that process.

    The Complexity of Modern Systems


    Modern computer systems have begun to rely on microservice architectures, patterns for high-availability and public cloud offerings. We build highly scalable systems that take advantage of ever-improving connectivity and the availability of third-party infrastructure and services. While the benefits are apparent, we find ourselves managing increasingly complex systems. Let’s consider the example of a web application that becomes unresponsive.

    Networking Problems

    Between the user and our web application, there is a complex web of connections running on private and public infrastructure. Service outages and hardware failures can easily affect a user’s ability to connect to our systems. Even within our systems, network calls travel between different instances and services. Our applications and services may be running perfectly, but if the connections between them are broken or experience performance degradation, the system won’t work as expected.

    Performance Issues

    Performance issues can occur due to hardware problems, constraints in frameworks and infrastructure we rely on, or poor design within the code itself. As the amount of traffic our applications handle increases, many of these problems grow exponentially, resulting in performance degradation or connection timeouts.

    Why Centralized Log Management is Essential

    Logging has always been a central component of understanding and troubleshooting applications. System and application logs provide an essential window into how data is processed and transferred, and the infrastructure's performance. Distributed systems complicate log management because there are so many different places where logs are created and stored. 

    Centralized log management solutions, like Mezmo, give you the ability to collect and aggregate all of your logs in a central location. Collection and aggregation are just part of the solution. Search capabilities allow you to find relevant logs to help troubleshoot production problems and monitor the health of different parts of your application. Many systems also include monitoring and configurable alerts to automatically identify problems and anomalies, and automatically alert support personnel to address them. 

    Comprehensive Alerting

    The most common types of alerts identify error conditions within your application. More advanced systems like Mezmo allow you to specify the type of error and its frequency as part of the alert. In addition to error states, you can also configure an alert based on the results of a predefined query against your logs.

    In addition to alerts based on the presence of an error condition or log query, you can also configure alerts based on the absence of certain types of logs. An example might be an online commerce site that expects a minimal level of transactions each hour. A lack of these transactions might indicate an access problem within the application.

    When used together, alerts for both the presence and absence of different conditions relieve your engineers of the responsibility to regularly review logs, and they can focus on building new features or improving their processes. You can learn more about alerting from the Mezmo Alerts Overview.

    Root Cause Analysis

    Once your support teams have received an alert from the log management system or a production incident has occurred, teams need to shift into resolution mode. The first step in the process is a root cause analysis. You can’t resolve a problem until you have identified the reason it occurred and each of the components involved.

    Identifying a problem in a distributed system can be incredibly challenging because you first need to identify which service is causing the problem and which services are affected by the situation. This investigation is where the ability to search through an aggregated collection of logs using request or transaction identifiers, time constraints, and additional filters is invaluable. You can learn more about how to search logs in the Mezmo Search Documentation.

    Once you’ve identified the problem and worked on a solution, you’ll typically want to deploy the solution while actively monitoring the logs. You can use the log management system to aggregate the logs, perform searches after time has passed, or take advantage of a helpful feature from Mezmo. Mezmo's Live Tail lets you monitor logs as they are received from your application, giving you real-time visibility into the new deployment status and allowing you to validate its success or failure.


    Learning More

    If you’d like to learn more about how Mezmo's centralized log management solution can work for you, there are a couple of different approaches you can take. You can schedule an appointment for a Mezmo team member to give you a live demo of how logging can enhance your production monitoring and support, and introduce you to some logging best practices. You can also sign up for a 14-day trial to try it for yourself. Speaking from personal experience, Mezmo is one of the most straightforward and most intuitive platforms available in the market today.

    false
    false
    Mike Mackrory

    5.27.21

    Mike Mackrory is a Global citizen who has settled down in the Pacific Northwest - for now. By day he works as an Engineer Manager for a DevOps team, and by night he writes and tinkers with other technology projects. When he's not tapping on the keys, he can be found trail-running, hiking and exploring both the urban and the rural landscape with his kids. Always happy to help out another developer, he has a definite preference for helping those who bring gifts of gourmet donuts, craft beer and/or Single-malt Scotch.

    SHARE ARTICLE

    RSS FEED