[SERVER-24943] High CPU usage fixed by server restart Created: 07/Jul/16  Updated: 23/Sep/16  Resolved: 23/Sep/16

Status: Closed
Project: Core Server
Component/s: None
Affects Version/s: 3.2.4
Fix Version/s: None

Type: Question Priority: Minor - P4
Reporter: udu Assignee: Kelsey Schubert
Resolution: Done Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified
Environment:

Debian Wheezy
12 CPUs, 24GB RAM on Google Compute


Attachments: File diagnostic.data.tar    
Participants:

 Description   

We noticed today that one of our Mongo servers had unexpectedly high CPU usage...which was fixed by restarting.

We were running about 258 ops/s on a Google Compute instance with 12 CPUs + 24GB RAM, with 30% CPU usage. I ended up restarting the instance, and we noticed the next set of ops was MUCH lower - it didn't rise above 3%.

We thought that was odd, so we repeated the test on another mongo server we had running the same code.
1) Run repeatable set of requests against mongo
2) Restart server
3) Run the same set of requests
4) Get drastically lower CPU post-restart.

Any ideas on what might be happening here?



 Comments   
Comment by Kelsey Schubert [ 23/Sep/16 ]

Hi developer@udu.nu,

Thank you for the update – I'm glad you were able to identify the root cause of the issue.

Kind regards,
Thomas

Comment by udu [ 23/Sep/16 ]

Hey Thomas,
It was a problem on our end; sorry about that.

Comment by Kelsey Schubert [ 08/Aug/16 ]

Hi developer@udu.nu,

Unfortunately, we have not been able to determine the cause of this behavior from the uploaded diagnostic.data. So we can continue to investigate, would you please reproduce this issue and provide the diagnostic.data, mongodb logs, and iostats from the run?

Please use the following script and upload iostat.log when the run completes:

delay=1
iostat -k -t -x ${delay:?} >iostat.log &

If you are able to provide a script that reproduces this issue or could provide additional details about your workload and dataset that would help us create a reproduction, it would greatly help our investigation.

Finally, there have been many improvements to WiredTiger since MongoDB 3.2.4, while I cannot point to particular ticket at this time, I would suggest that you consider upgrading to the latest version of MongoDB 3.2 and confirming whether this issue persists.

Thank you,
Thomas

Comment by udu [ 18/Jul/16 ]

Any update on this?

Comment by udu [ 07/Jul/16 ]

Hey Thomas,

Attached it.

Cheers,
Bruce

Comment by Kelsey Schubert [ 07/Jul/16 ]

Hi developer@udu.nu,

Thanks for the report. So we can continue to investigate this behavior, would you please archive (tar or zip) the $dbpath/diagnostic.data directory and attach it to this ticket?

Kind regards,
Thomas

Generated at Thu Feb 08 04:07:50 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.