How Sentry Helps Grofers Meet Unprecedented Delivery Demand
As one of the leading grocery delivery services in India, Grofers needed to make massive fulfillment adjustments as demand rocketed due to an entire nation sheltering in place to help control the spread of infection during a global pandemic.
For the Grofers team, providing much-needed grocery delivery service is as much of a technical challenge as a human resources one. Sentry error monitoring has become an integral part of Grofer’s development process, helping the engineering team scale and work efficiently as consumer demand has spiked.
For VP of Engineering Vaidik Kapoor, Sentry has helped Grofers future proof the business by setting up a development team that could scale with these unprecedented global changes. “We’re seeing demand like no one expected because we’re experiencing a global pandemic that no one expected. Sentry is one of the main factors that helps us meet this demand and deliver quality service and supplies in such uncertain times,” explained Vaidik.
Sentry is one of the main factors that helps us meet this demand and deliver quality service and supplies in such uncertain times.
It wasn’t always this way. Before Sentry, Grofers’ engineers lacked the proper tools to efficiently monitor errors in production. Vaidik’s team conducted standard monitoring using web server metrics (like 5XX), but to dig any deeper, they needed to manually review logs. Monitoring became such a problem that some of the teams tried to build a homegrown workaround. The team tried to build monitoring tools with global try/except blocks that would catch the error and send an email with the stack trace to the entire engineering team. Unfortunately, they were unable to capture enough context from these emails to actually debug the issue.
Choosing Sentry and the Python Ecosystem
Vaidik knew his team needed a robust error monitoring platform, but tools that could provide the level of service he needed were hard to find. Vaidik knew several developers who used Sentry in the past, so he decided to explore further.
Vaidik found Sentry’s extensive Python error tracking capabilities a huge benefit to Grofers, which uses Python extensively across the application. Better yet, many libraries in the Python ecosystem synced well with Sentry, plus additional libraries were built directly on top of Sentry. Since Sentry was open-source, Vaidik knew he was choosing a service that wouldn’t lock them out.
Trimming Debug Time From Days to Minutes
Before Grofers onboarded Sentry, the team burned from 4 to 8 hours per developer each week debugging technical issues. In addition to developer productivity loss, these unresolved errors caused serious customer experience problems that often stood unresolved for days and months. Some of the performance issues took weeks to resolve due to a lack of visibility, creating negative customer sentiment.
After Grofers onboarded Sentry they improved developer productivity by 30% as developers gained more visibility around code exceptions. Vaidik explained that once Sentry was fully implemented, “The time to resolve errors went from days to minutes.”
The time to resolve errors went from days to minutes.
For the past 5 years, Sentry has served as an integral part of Grofers code monitoring process. The moment an engineer identifies an error rate increase, the first thing they look at is Sentry.
While Vaidik’s team has invested in the logging infrastructure, they don’t have to hunt through logs with Sentry. As soon as monitoring systems detect any SLO breaches, they look at Sentry first as it gives them so much more context to reproduce and quickly resolve the issue. In most cases, they don’t have to go all the way to find logs, which saves time and improves customer satisfaction.
With Sentry, Grofers can scale development to meet unprecedented demand in an unprecedented time in human history. Follow their tech journey at Lambda by Grofers to learn more.
Developers don’t have to dig through logs or guess why code may have broken. They can build, iterate, and grow.