Bug 2255436 - [RHCS 5] RFE: change default value of "mds_bal_interval" to "0", aka false
Summary: [RHCS 5] RFE: change default value of "mds_bal_interval" to "0", aka false
Keywords:
Status: CLOSED ERRATA
Alias: None
Product: Red Hat Ceph Storage
Classification: Red Hat Storage
Component: CephFS
Version: 5.3
Hardware: All
OS: All
unspecified
medium
Target Milestone: ---
: 5.3z6
Assignee: Patrick Donnelly
QA Contact: Hemanth Kumar
Ranjini M N
URL:
Whiteboard:
Depends On: 2227309 2255435
Blocks: 2258797
TreeView+ depends on / blocked
 
Reported: 2023-12-20 18:37 UTC by Patrick Donnelly
Modified: 2024-06-02 02:28 UTC (History)
8 users (show)

Fixed In Version: ceph-16.2.10-231.el8cp
Doc Type: Enhancement
Doc Text:
.The MDS default balancer is now disabled by default With this release, the MDS default balancer or the automatic dynamic subtree balancer is disabled by default. This prevents accidental subtree migrations, Subtree migrations can be expensive to undo when the operator increases the file system `max_mds` setting without planning subtree delegations, such as, with pinning.
Clone Of:
Environment:
Last Closed: 2024-02-08 16:57:20 UTC
Embargoed:


Attachments (Terms of Use)


Links
System ID Private Priority Status Summary Last Updated
Ceph Project Bug Tracker 61378 0 None None None 2023-12-20 18:37:18 UTC
Red Hat Issue Tracker RHCEPH-8110 0 None None None 2023-12-20 18:37:33 UTC
Red Hat Knowledge Base (Solution) 7026124 0 None None None 2024-06-02 02:28:21 UTC
Red Hat Product Errata RHSA-2024:0745 0 None None None 2024-02-08 16:57:23 UTC

Description Patrick Donnelly 2023-12-20 18:37:18 UTC
This bug was initially created as a copy of Bug #2227309

I am copying this bug because: 

rhcs 5 backport

Description of problem:  RFE: change default value of "mds_bal_interval" to "0", aka false

From case 03492882 and BZ (https://bugzilla.redhat.com/show_bug.cgi?id=2203258) we see that having "mds_bal_interval" enabled results in performance issue. Beyond the log evidence in BZ 2203258, the customer reports that Ceph FS latency has dropped by 75 percent and I/O in the cluster has doubled since "mds_bal_interval" was set to false on their system.

We'd like to see the change in 5.3z-whatever and 6.1z-whatever and beyond

See also KCS #7026124, (https://access.redhat.com/solutions/7026124)

This same customer commented if this feature hampers performance this much and there is no intention on fixing it, the default value should be changed.


Version-Release number of selected component (if applicable):


How reproducible:


Steps to Reproduce:
1.
2.
3.

Actual results:


Expected results:


Additional info:

Comment 9 errata-xmlrpc 2024-02-08 16:57:20 UTC
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory (Moderate: Red Hat Ceph Storage 5.3 Security update), and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHSA-2024:0745


Note You need to log in before you can comment on or make changes to this bug.