Project

General

Profile

Actions

Feature #40562

open

More granular setting for alerting full ratios for disks with different sizes

Added by Neha Ojha almost 5 years ago.

Status:
New
Priority:
Normal
Assignee:
-
Category:
-
Target version:
-
% Done:

0%

Source:
Tags:
Backport:
Reviewed:
Affected Versions:
Component(RADOS):
Pull request ID:

Description

Following is the RFE description:

When we're using heterogeneous hardware i.e. disks having different sizes.
There's no provision to set the values of the nearfull and full ratio per OSD level.

For example if we're using a 1T disk, it will report at .85T or ~150G remaining.
And when we're using a 12T disk , the nearfull ratio of 80% on the 12T means they will report at ~9.6T, or 2.4T remaining. <<<< This value is still high

Similarly for full ratio ,
For example if we're using a 1T disk, it will report at .95T or ~50G remaining.
And when we're using a 12T disk , the full ratio of 95% on the 12T means they will report at ~11.4T, or 600G remaining. <<<< The value is still high

Its a concern and should be taken into consideration when we want to utilize more of the remaining disk space as the disk size grows.
The values of mon_osd_full_ratio, mon_osd_nearfull_ratio and osd_backfill_full_ratio should be taken into consideration.

We can have more granular setting for alerting full ratios :
- Per crush bucket / root level ; If we tend to have the same sized drives in the same crush bucket
- OSD level

No data to display

Actions

Also available in: Atom PDF