Re: And again about memory leaks

[Dwight Merriman]

check # of connectiosn to the DB

there is a thread/stack per connection.  if a huge # that would be one explanation.

On Thursday, November 15, 2012 2:40:32 AM UTC-5, Vladimir Yakovlev wrote:

Hi Guys,

We deployed a new server yesterday, based on latest CentOS (6.3) and MongoDB 2.0.7, running in a ReplicaSet (role Secondary).

It took a while to sync between Primary and Secondary, but afterwards we began to receive messages like the following:

Thu Nov 15 10:32:19 [conn103911] warning: virtual size (82820MB) - mapped size (74014MB) is large (8587MB). could indicate a memory leak

And if one runs a db.serverStatus().mem it returns the following:

{

        "bits" : 64,

        "resident" : 141,

        "virtual" : 82599,

        "supported" : true,

        "mapped" : 37007,

        "mappedWithJournal" : 74014,

        "note" : "virtual minus mapped is large. could indicate a memory leak"

}

There is a total of 24GB RAM, nothing else but MongoDB on this machine.

I saw some posts with similar issues, but didn't find any suitable solution for us, but threads like "this is not a serious issue, if it disappears in a while".

Still, any ideas, what could be done to avoid the precedent?

Thanks in advance,

BR,

Vladimir.