[SERVER-13224] User defined steps to handle out of disk conditions Created: 17/Mar/14  Updated: 06/Dec/22  Resolved: 24/Aug/18

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

Type: Improvement Priority: Major - P3
Reporter: Nick Pellant Assignee: Backlog - Storage Execution Team
Resolution: Won't Fix Votes: 3
Labels: nyc
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Related
is related to SERVER-3759 filesystem ops may cause termination ... Closed
Assigned Teams:
Storage Execution
Participants:

 Description   

When the MongoDB's dbPath has run out of space, no operations are allowed at the MongoD level, such as reads, and new connections are blocked.

This situation can lead to downtime until the dbPath 's volume is increased in space or a new disk is attached to the system.

It would be nice to allow the user to have rules in place which would change the behaviour of the MongoD process in the event of an out of disk condition. The user should be allowed to run administrative actions on the MongoD process from the shell. These include deleting databases, collections and documents, as well as attempting a compact.

These rules, or policies, that the user passes to the MongoD process in advance of this out of disk condition could be something like:

  1. Enable noprealloc
  2. Enable the use of smallfiles
  3. Auto restart the MongoD process, on a different port and with the replSetname ignored or removed
  4. Preform a compaction, drop a database, etc.

Naturally, this should be logged with a relevant message of what was actually done.



 Comments   
Comment by Ian Whalen (Inactive) [ 24/Aug/18 ]

We don't plan to make this a configurable option but instead will clarify out of disk behavior in one direction. asya to link any tickets here about which direction we plan to go.

Generated at Thu Feb 08 03:31:02 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.