OPERATIONAL DEFECT DATABASE
...
...
BugZero found this defect 2765 days ago.
When few queries process many documents (for example, full collection scan), MongoDB allocate and deallocate too much memory. So after few full scan query, free memory of linux is decreased and eventually linux use swap space. Take a look at the attached graph. And this case happen very often. I think you can reproduce this case easily.
mark.agarunov commented on Wed, 5 Jul 2017 15:14:18 +0000: Hello matt.lee, Thank you for the response. As this issue cannot currently be reproduced, I've closed this ticket. If this is an issue for you in the future, please let us know and we will reopen the ticket. Thanks, Mark sunguck.lee@gmail.com commented on Fri, 30 Jun 2017 08:50:20 +0000: Hi Mark. I have seen this case in quite a while, so diagnostics data is gone. So I have tried to reproduce it but I can't. I will get back again with diagnostics data when I see similar case. Thanks. mark.agarunov commented on Thu, 22 Jun 2017 15:34:20 +0000: Hello matt.lee, Thank you for the report. To better understand what may be causing the memory usage, could you please archive and upload the $dbPath/diagnostic.data directory and attach it to this ticket? This should give us a better idea of what may be using the memory. Thanks, Mark
1) Insert 50000000 documents in the test collection. 2) Query all documents with several client threads 3) Check linux memory usage
