yCrash is our end-end root cause analysis solution built on top of our successful products GCeasy, fastThread, HeapHero. yCrash leverages the same underlying parsing engine and ML algorithms of these 3 tools. On top of it, yCrash provides additional capabilities. This document intends to highlight those key capabilities.
Quick Overview
In this video clip. our architect talks about the key features and gives a demo of yCrash product.
Capabilities
Below are key additional capabilities that are present in yCrash (that aren’t available in GCeasy, FastThread, HeapHero):
- Predicting outages: yCrash monitors micro-metrics and predicts the outages before it surfaces in the production environment. Examples 45 minutes before OutOfMemoryError surfaces in the production environment, yCrash can alert you.
- 360-degree data capturing: To troubleshoot production problems effectively, you need to capture essential data. Lot of times, because of the heat of moment, application are restarted without capturing essential data. To address this issue, yCrash captures 360-degree data. It captures garbage collection log, thread dump, heap dump, netstat, vmstat, top, top -H, iostat, kernel parameters, application log, kernel log, ping, … to effectively troubleshoot the problem.
- Instant RCA: yCrash automatically captures the above mentioned 360-degree artifacts right away when the problem surfaces in production environment. Instantly after capturing these artifacts, yCrash analyzes them and generates one unified root cause analysis report.
- Data Marriage: GCeasy, FastThread, HeapHero tools would analyze the dumps in silos. i.e., GCeasy tool only analyzes garbage collection logs. FastThread only analyzes thread dumps. HeapHero tool only analyzes the heap dumps. yCrash analyzes the data in silo and marries the data of these silo analysis to provide holistic information. yCrash marries thread dump data with top -H output to identify the exact line of code causing the CPU to spike. It marries heap dump data with thread dump data to see what threads and lines of code that cause the memory leak. It marries ‘top’ output with ‘ps’ output, …
- Security: As a support engineer, you might have challenges transmitting the captured dump files from your device to your developers/SME to debug the problem. You might have to upload it to an FTP site within your organization if it’s available. Sometimes we have seen support engineers uploading these dump files to SharePoint, Wiki, and even external drives like google drive, dropbox due to lack of FTP site availability. As you are aware, these dump files tend to contain sensitive information such as: IP addresses, code path, customer SSN, credit card numbers….. Once you have uploaded the data to FTP site, developers will have to download the dumps to their local laptops for analysis. Now, these sensitive dumps are distributed in multiple locations opening up the room for various security vulnerabilities. In yCrash all dumps are securely transmitted from your device (through https) and stored in one place, i.e. yCrash server. Users only see the analyzed results through their browser. No one will have access to the raw data.
- Integration with tools: yCrash integrates with all major monitoring tools such as AppDynamics, Elastic Search, Grafana, Promethus,.. yCrash also integrates with major ITSM systems like JIRA, ServiceNow.
- Dashboard & Historical archival: yCrash provides a dashboard (which is not available in earlier 3 products). From the dashboard, you can see all production incidents reported in your enterprise from one single view. You can go back in time (like last month, last year) and see all the incidents in your enterprise.
- Comparison tool: In GCeasy, fastThread, HeapHero you can’t compare the metrics, statistics, graphs between two reports. But in yCrash you can do so. This comparison is especially useful when you have gone live with a new release. You might want to compare what was your GC throughput, pause times,… between your latest release and old release.
- Search: yCrash provides you with a search capability that is not available in other 3 tools. You can search the reports based on the host name, IP addresses, date range, application name, ….
Installation
yCrash server installation is the same as the other 3 products. Just unzip and run it.
But for automatic data capturing, you need to install the yCrash agents (which is a self-contained GOlang script) on all your devices. If you prefer not to install our agents on your devices, there are two options:
- You can build your own agent using our simple REST APIs.
- You can upload the dumps manually from the yCrash dashboard
Leave a Reply