Hello, everyone. I’m new to the MongoDB family and still have limited knowledge about MongoDB. I would appreciate your assistance.
I have been dealing with an unusual performance issue in MongoDB. To investigate the root cause of the performance problem, I used sysdig to record the system conditions during the time of the issue. The problem occurs every Saturday around 11:00 AM, when disk I/O becomes highly busy, reaching over 100%. Since the default logs didn’t provide any insights into the issue, I resorted to using sysdig to capture the system conditions during the faulty period. Initially, I wanted to examine system calls like pwrite or write to see what data was being written. However, I realized that the data written to WiredTiger (wt) is compressed or encrypted and cannot be directly read. Is there a way to convert this text into readable content? Additionally, I’m curious about the compression or encryption algorithms used in this process.
Hey @wu_feng1,
Welcome to the MongoDB Community!
Since you’re seeing high disk I/O utilization around the same time every Saturday, is it possible that you have a batch process that happens at that time regularly?
Few things to check:
-
Could you confirm if you have any such cron jobs, scripts, or other periodic tasks configured to run around the specified time interval? Seeing what jobs overlap with the performance spike could help identify the cause.
-
Could you share the output of
mongostatand the MongoDB server logs of that particular timeframe? -
Please provide the MongoDB version, disk size, and deployment configuration. Also, are you utilizing Docker, a VM, or a similar solution to host your MongoDB server?
Looking forward to your response.
Best regards,
Kushagra