Red Hat Bugzilla – Bug 1201740
[New] - Adding RHS 3.0.4 node to RHSC overrides the values of vm.dirty_ratio and dirty_background_ratio to 5 and 2 from RHEL defaults.
Last modified: 2015-07-29 01:27:26 EDT
Description of problem: As of now rhsc host-deploy overrides the existing tuned profile on RHS nodes and sets it to rhs-virtualization and because of this the values of vm.dirty_ratio and dirty_background_ratio changes to 5 and 2 and does not remain the RHEL default values 20 and 10. UI should provide an option where user is able to select the profile which he wants to set on the node. Version-Release number of selected component (if applicable): rhsc-3.0.4-1.25.el6rhs.noarch How reproducible: Always. Steps to Reproduce: 1. Install RHS 3.0.4 ISO. 2. Now add the node to rhsc. 3. Actual results: host-deploy overrides the existing profile to rhs-virtualization and because of this vm.dirty_ratio and dirty_background_ratio changes to 5 and 2 and there might be some performance impact because of this. Expected results: Either UI should provide an option where user is able to select the profile which he wants to set on the node. Additional info:
Ramesh, Can you please review the edited doc text and clear the needinfo flag?
Doc text looks good to me
Currently by default rhs-virtualization profile is set for RHS nodes. One approach is to change this default to rhs-high-throughput profile, which is recommended for RHS nodes. The other option is to ask user to choose the profile during cluster creation in RHSC, and set the profile accordingly.
As discussed in triage meeting, the solution for this issue; User would be provided an option to choose the tuned profile during cluster creation.
Please provide the fixed in version.
Created a cluster with gluster Tuned Profile : rhs-high-throughput and added a RHS 3.0.4 ISO to the cluster. I see the following as the profile in the RHS node which is added to RHSC and vm_dirty_background and vm_dirt)_ [root@dhcp37-153 ~]# tuned-adm list Available profiles: - server-powersave - spindown-disk - desktop-powersave - virtual-guest - virtual-host - rhs-virtualization - rhs-high-throughput - enterprise-storage - laptop-ac-powersave - laptop-battery-powersave - default - throughput-performance - latency-performance Current active profile: rhs-high-throughput [root@dhcp37-153 ~]# sysctl -a | grep "dirty" vm.dirty_background_ratio = 10 vm.dirty_background_bytes = 0 vm.dirty_ratio = 20 vm.dirty_bytes = 0 vm.dirty_writeback_centisecs = 500 vm.dirty_expire_centisecs = 3000 When the cluster is created with rhs-virtualization profile and adds a node to this cluster the following values are seen. [root@dhcp37-153 ~]# tuned-adm list Available profiles: - server-powersave - spindown-disk - desktop-powersave - virtual-guest - virtual-host - rhs-virtualization - rhs-high-throughput - enterprise-storage - laptop-ac-powersave - laptop-battery-powersave - default - throughput-performance - latency-performance Current active profile: rhs-virtualization [root@dhcp37-153 ~]# sysctl -a | grep "dirty" vm.dirty_background_ratio = 1 vm.dirty_background_bytes = 0 vm.dirty_ratio = 20 vm.dirty_bytes = 0 vm.dirty_writeback_centisecs = 500 vm.dirty_expire_centisecs = 3000 setting the profile as rhs-virtualization changes the vm.dbr to 1 and leaves the vm.dr to what ever it was before.
Verified and works fine with rhsc-3.1.0-0.57.master.el6.noarch
Darshan, Please review and sign-off the edited doc text.
(In reply to Divya from comment #11) > Darshan, > > Please review and sign-off the edited doc text. Looks Good.
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, 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://rhn.redhat.com/errata/RHEA-2015-1494.html