Project

General

Profile

Feature #20298

store longer dup op information

Added by Josh Durgin almost 7 years ago. Updated almost 6 years ago.

Status:
Resolved
Priority:
High
Category:
-
Target version:
-
% Done:

0%

Source:
Tags:
Backport:
jewel
Reviewed:
Affected Versions:
Pull request ID:

Description

Under some workloads, it is beneficial to avoid log-based recovery and use backfill instead. This lets I/O proceed on objects being backfilled, since it occurs in the background. This is implemented by setting the min/max pg log size to 1 and 2.

This has a side effect of eliminating our duplicate op detection. To make this configuration robust, we need to keep a separate structure to detect duplicate ops beyond the pg log entries.


Related issues

Blocked by Ceph - Bug #21026: PR #16172 causing performance regression Resolved 08/17/2017
Copied to RADOS - Backport #22405: jewel: store longer dup op information Rejected

History

#1 Updated by Josh Durgin over 6 years ago

  • Status changed from In Progress to Resolved
  • Backport deleted (jewel, kraken)

#2 Updated by Ken Dreyer over 6 years ago

  • Status changed from Resolved to Pending Backport
  • Backport set to jewel

#3 Updated by Ken Dreyer over 6 years ago

  • Blocked by Bug #21026: PR #16172 causing performance regression added

#4 Updated by Nathan Cutler over 6 years ago

#5 Updated by J. Eric Ivancich over 6 years ago

PR Created: https://github.com/ceph/ceph/pull/19497

Includes Josh Durgin's fix for performance regression.

#6 Updated by Nathan Cutler almost 6 years ago

  • Status changed from Pending Backport to Resolved

Also available in: Atom PDF