Project

General

Profile

Actions
Copy link

Enhancement #71764

open

Support devices larger than 100 TiB

Added by Dan van der Ster 8 months ago. Updated 8 months ago.

Status:
Fix Under Review
Priority:
Normal
Assignee:
Dan van der Ster
Category:
-
Target version:
-
% Done:

0%

Source:
Backport:
squid,tentacle
Reviewed:
Affected Versions:
ceph-qa-suite:
Component(RADOS):
Pull request ID:
64056
Tags (freeform):
Merge Commit:
Fixed In:
Released In:
Upkeep Timestamp:

Description

We currently don't allow devices to have a CRUSH weight larger than 100.0, ergo 100 TiB using our current device-size / CRUSH-weight equivalence:

src/crush/crush.h:#define CRUSH_MAX_DEVICE_WEIGHT (100u * 0x10000u)

120 TiB devices are already on the market.

The fix for #71763 will help here too, but IMHO we should just increase CRUSH_MAX_DEVICE_WEIGHT to 1000u or more.
I don't think 100u is a special value in the CRUSH algorithm.

We need to update the kernel too, even though it isn't used anywhere:

include/linux/crush/crush.h:#define CRUSH_MAX_DEVICE_WEIGHT (100u * 0x10000u)

Related issues 1 (1 open — 0 closed)

Related to RADOS - Enhancement #71763: Support clusters larger than 65.5 petabytesFix Under ReviewDan van der Ster

Actions
Actions
Copy link
 #1

Updated by Dan van der Ster 8 months ago

Actions
Copy link
 #2

Updated by Dan van der Ster 8 months ago

  • Pull request ID set to 64056
Actions
Copy link
 #3

Updated by Dan van der Ster 8 months ago

  • Pull request ID changed from 64056 to 64055
Actions
Copy link
 #4

Updated by Dan van der Ster 8 months ago

  • Pull request ID changed from 64055 to 64056
Actions
Copy link
 #5

Updated by Dan van der Ster 8 months ago

  • Backport set to squid,tentacle
Actions
Copy link
 #6

Updated by Dan van der Ster 8 months ago

  • Status changed from New to Fix Under Review
Actions
Copy link

Also available in: Atom PDF