[SERVER-27315] Mongo wt cache will not be reclaimed Created: 07/Dec/16  Updated: 27/Oct/23  Resolved: 13/Dec/16

Status: Closed
Project: Core Server
Component/s: WiredTiger
Affects Version/s: 3.2.9
Fix Version/s: None

Type: Bug Priority: Major - P3
Reporter: liangzhang Assignee: Kelsey Schubert
Resolution: Works as Designed Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified
Environment:

CentOS release 6.6


Attachments: PNG File QQ图片20161214132506.png     PNG File Screen Shot 2016-12-12 at 9.50.42 am.png     File diagnostic.data.tgz     PNG File page evicted.png     PNG File page passed.png     PNG File page split.png     PNG File wt_cache.png    
Participants:

 Description   

At 18:15 this time point, wt cache continues to grow until 96%. then at 19:00, we switch primary to secondary.
Since then, this mongo service almost without any request, but the wt cache is not recovered, has remained at around 92%.
I guess some reason is that the cache can not be recycled.

System environment
System: CentOS release 6.6
Kernel: 2.6.32-504.el6.x86_64
Memory: 256G
Disk: SSD-PCIE
File System: xfs
mongo version: 3.2.9
Config:

Storage:
DbPath: "/ mongo / db"
DirectoryPerDB: true
Engine: wiredTiger
WiredTiger:
EngineConfig:
CacheSizeGB: 200
JournalCompressor: none
CollectionConfig:
BlockCompressor: snappy
Security:
KeyFile: "/ mongo / mongodb-keyfile"



 Comments   
Comment by Kelsey Schubert [ 14/Dec/16 ]

Hi ufozhangliang@163.com,

Please note that SERVER project is for reporting bugs or feature suggestions for the MongoDB server. For MongoDB-related support discussion please post on the mongodb-users group or Stack Overflow with the mongodb tag. A question like this involving more discussion would be best posted on the mongodb-users group.

Kind regards,
Thomas

Comment by liangzhang [ 14/Dec/16 ]

What causes eviction to be so busy. Is a lot of update, find, findandmodify or any other reason? How can I optimize to reduce the eviction workload?

Comment by Kelsey Schubert [ 13/Dec/16 ]

Hi ufozhangliang@163.com,

We've examined the diagnostic.data you've provided. It appears that even after the node steps down to secondary it is still under considerable load. Additionally, we see that eviction is busy working under this workload. In this case, it is expected that the secondary would continue to use cache and CPU resources.

Kind regards,
Thomas

Comment by liangzhang [ 08/Dec/16 ]

Thank you for your reply, I hope you can help me solve my problems.

Comment by Kelsey Schubert [ 07/Dec/16 ]

Hi ufozhangliang@163.com,

Would you please attach an archive of the diagnostic.data directory to this ticket so we can continue to investigate?

Thank you,
Thomas

Generated at Thu Feb 08 04:14:49 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.