0

We're running a MongoDB 3.0 server with multiple users accessing it and we're experiencing very frequent disk reads by mongod process utilizing our SSD drives to 100%

During these peaks the page faults increase by 30x - it seems like a query is accessing data not in memory which in turn needs to be loaded.

I couldn't find a way how to monitor which query/collection/db triggers these reads. Any ideas if it's even possible with MongoDB?

All we can see is that iotop reports 10-15s disk read at full speed from mongod process.

3
  • 1) What is the RAM usage of your VPS? If it is running peak, you probably need to upgrade it. 2) WHat is the disk I/O when you say it is running 100%. Some provider limits IOPs which turns into lower disk I/O performance. 3) If above are fine, mongoperf and mongotop will be the option for you to start troubleshooting with docs.mongodb.com/v2.2/reference/mongotop docs.mongodb.com/manual/reference/program/mongoperf May 27, 2016 at 23:11
  • 1) 60% usage, still 2GB RAM available 2) we get 3000 iops 3) we've tried the tools, however they're showing only aggregate data, not per collection or db which makes it hard to find out which user is triggering the queries
    – mike
    May 28, 2016 at 7:56
  • @MukeshJagani You should post that comment as an answer, because that is exactly how this should be troubleshot. Sep 9, 2016 at 10:57

0

You must log in to answer this question.

Browse other questions tagged .