[SERVER-9285] Very high usage of memory on upgrading from 2.2 to 2.4.1 Created: 08/Apr/13  Updated: 10/Dec/14  Resolved: 04/Apr/14

Status: Closed
Project: Core Server
Component/s: Performance
Affects Version/s: 2.4.1
Fix Version/s: None

Type: Bug Priority: Major - P3
Reporter: Ronak Kumar Samantray Assignee: Unassigned
Resolution: Done Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified
Environment:

Windows Server 2008 R2


Operating System: ALL
Participants:

 Description   

After upgrading from 2.2 to 2.4.1 i am seeing a very usage of the memory. The memory (8GB) usage stays at 40-50% most of the time and gradually goes upto 98% !!

Please help urgently as it is impacting my production



 Comments   
Comment by Tad Marshall [ 12/Apr/13 ]

The release candidate version 2.4.2-rc0 includes fixes for some memory leaks in JavaScript, which are particularly evident when running MapReduce but also affect other JavaScript usage. Can you try version 2.4.2-rc0 and see if the memory usage is better?

Comment by Ronak Kumar Samantray [ 12/Apr/13 ]

I just reverted back to 2.2.* and its working fine to start with. I will wait for a full day and let you guys know. But as of now it looks like a problem with the 2.4.1 version.

Someone kindly help us resolve this.

Ronak

Comment by Steinn Steinsen [ 10/Apr/13 ]

Hi,

I am running a mongod server and am experiencing the same problem running on Ubuntu 12.04.

I have a replica set with two nodes and an arbiter. This issue has occurred at least three times since upgrading from 2.2.3 to 2.4.1.
This has only occurred on the primary node. The secondary node is only used for fail overs. There are no MapReduce jobs, so it's probably not
the 9230 issue.

Currently when the primary node gets to high memory usage I step down the primary and restart the mongod process.

Any idea what the problem might be?

Comment by Ronak Kumar Samantray [ 09/Apr/13 ]

I do have Map-Reduce functions, but its only used very few times. The primary queries are Read and Geo spatial queries.

Comment by Tad Marshall [ 09/Apr/13 ]

Do you run MapReduce jobs?

This may be SERVER-9230 (which will be fixed in 2.4.2), in which a memory leak causes heap space to grow over time. Can you run VMMap (http://technet.microsoft.com/en-us/sysinternals/dd535533.aspx) on the mongod.exe process and see if the heap memory is increasing over time?

Comment by Ronak Kumar Samantray [ 09/Apr/13 ]

I am observing it on mongod.exe

Comment by Randolph Tan [ 08/Apr/13 ]

Hi,

Are you seeing this on mongos or mongod?

Generated at Thu Feb 08 03:19:57 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.