[SERVER-15025] Disk usage increases linearly with constant data size, large churn, usePowerOf2Sizes and large document sizes Created: 25/Aug/14  Updated: 10/Dec/14  Resolved: 26/Aug/14

Status: Closed
Project: Core Server
Component/s: Storage
Affects Version/s: 2.4.6
Fix Version/s: None

Type: Bug Priority: Critical - P2
Reporter: Alan Spencer Assignee: Unassigned
Resolution: Duplicate Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Attachments: File simple_chunk_store_mod_repro.py    
Issue Links:
Depends
Duplicate
duplicates SERVER-14088 Excessive DB fileSize growth for larg... Closed
Related
Operating System: ALL
Steps To Reproduce:

A reproduction script will be attached.

Participants:

 Description   

MongoDB will erroneously allocate new space at essentially a constant rate for the following:

  • An insert/delete only workload
  • usePowerOf2Sizes used
  • essentially constant, at least bounded data size
  • Documents greater than 2MB.

A common use case where this applies is where a collection represents, for example, the last 48 hours of some data. Over the 48 hours data is constantly inserted, at some point old data is deleted in bulk.

This results in a constant data size resulting in a linearly increasing database on disk.



 Comments   
Comment by Alan Spencer [ 26/Aug/14 ]

Just to ensure this doesn't get lots in duplicate tagging - this issue is for 2.4.6 - not 2.6 with the limited scanning of MAX_BUCKET, it is not caused by that change (though on 2.6 is it extremely exaggerated)

Should we leave this open as 2.4 specific?

Comment by Eliot Horowitz (Inactive) [ 26/Aug/14 ]

Its actually a dup of SERVER-14088, which is different than 2958.

Comment by Scott Hernandez (Inactive) [ 25/Aug/14 ]

How is this not a duplicate of SERVER-2958 ?

Comment by Alan Spencer [ 25/Aug/14 ]

Reproduction script

Generated at Thu Feb 08 03:36:42 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.