Note: This bug is displayed in read-only format because
the product is no longer active in Red Hat Bugzilla.
RHEL Engineering is moving the tracking of its product development work on RHEL 6 through RHEL 9 to Red Hat Jira (issues.redhat.com). If you're a Red Hat customer, please continue to file support cases via the Red Hat customer portal. If you're not, please head to the "RHEL project" in Red Hat Jira and file new tickets here. Individual Bugzilla bugs in the statuses "NEW", "ASSIGNED", and "POST" are being migrated throughout September 2023. Bugs of Red Hat partners with an assigned Engineering Partner Manager (EPM) are migrated in late September as per pre-agreed dates. Bugs against components "kernel", "kernel-rt", and "kpatch" are only migrated if still in "NEW" or "ASSIGNED". If you cannot log in to RH Jira, please consult article #7032570. That failing, please send an e-mail to the RH Jira admins at rh-issues@redhat.com to troubleshoot your issue as a user management inquiry. The email creates a ServiceNow ticket with Red Hat. Individual Bugzilla bugs that are migrated will be moved to status "CLOSED", resolution "MIGRATED", and set with "MigratedToJIRA" in "Keywords". The link to the successor Jira issue will be found under "Links", have a little "two-footprint" icon next to it, and direct you to the "RHEL project" in Red Hat Jira (issue links are of type "https://issues.redhat.com/browse/RHEL-XXXX", where "X" is a digit). This same link will be available in a blue banner at the top of the page informing you that that bug has been migrated.
Description of problem: Value of parameters set in throughput-performance profile are no longer appropriate for best performance of sap-hana. While working on a certification performance issue, the following settings were found to improve performance.
Kernel.sched_min_granularity_ns from 10000000 to 3000000
Kernel.sched_wakeup_granularity_ns from 15000000 to 4000000
The 10000000 / 15000000 settings are currently inherited from throughput-performance.
We also found that readahead parameter was better left at the default of 128K rather than the 4096K that is set in throughput-performance.
One performance investigation is outlined in https://bugzilla.redhat.com/show_bug.cgi?id=1745111 for Power9, another example is in https://bugzilla.redhat.com/show_bug.cgi?id=1753205 for x86_64.
Version-Release number of selected component (if applicable):
tuned-2.13.0-0.1.rc1.el8.noarch
How reproducible:
Every time
Steps to Reproduce:
1. Load SAP workload fsperf and scsi based I/O
2. Run 4K random overwrite test - once with lower sched_ settings, change to higher settings and run again
3. Compare results
Actual results: Results don't meet SAP KPIs with higher (throughput-performance based) sched_ settings
Expected results: Results meet and exceed SAP KPIs with lower sched_settings
Additional info: Recommending no longer inheriting throughput-performance profile and putting required settings directly in sap-hana profile
This is a RHEL8.2 blocker request as it is gating an SAP cert. The testing for this is also being requested for the low_latency tuned in 8.2
Its been tested very well on by both SAP workloads and with the Perf team / kernel QE perf team.
Comment 5Jaroslav Škarvada
2019-12-06 15:47:01 UTC
Hi Jaroslav,
There are a few other settings that we do need to carry over from throughput-performance profile into the new sap-hana profile. The [cpu] settings are necessary from a performance perspective.
The vm. parameters have been set for all the tests that we have done so would like to include them as well.
[cpu]
governor=performance
energy_perf_bias=performance
min_perf_pct=100
[sysctl]
vm.dirty_ratio = 40
vm.dirty_background_ratio = 10
vm.swappiness=10
Thanks!
Dave
Comment 8Jaroslav Škarvada
2019-12-06 16:40:56 UTC
(In reply to djdumas from comment #7)
> Hi Jaroslav,
>
> There are a few other settings that we do need to carry over from
> throughput-performance profile into the new sap-hana profile. The [cpu]
> settings are necessary from a performance perspective.
> The vm. parameters have been set for all the tests that we have done so
> would like to include them as well.
>
>
> [cpu]
> governor=performance
> energy_perf_bias=performance
> min_perf_pct=100
>
> [sysctl]
> vm.dirty_ratio = 40
> vm.dirty_background_ratio = 10
> vm.swappiness=10
>
>
> Thanks!
> Dave
Hi Dave,
please check the PR from the comment 5 (direct link: https://github.com/redhat-performance/tuned/pull/231/commits/52156a00b590ef401b38dcc65720b344d67e86f6) whether it's OK for you. It's the single tuning for all architectures.
It's simple profile, so I directly copied the settings from the throughput-performance profile to the sap-hana not to complicate things, but if the following works better for you, I could change it, e.g.:
- set readahead explicitly to the 128kB (now it's left on its defaults)
- keep throughput-performance included, and just override what's needed in sap-hana
Jaroslav
(In reply to Jaroslav Škarvada from comment #8)
>
> Hi Dave,
>
> please check the PR from the comment 5 (direct link:
> https://github.com/redhat-performance/tuned/pull/231/commits/
> 52156a00b590ef401b38dcc65720b344d67e86f6) whether it's OK for you. It's the
> single tuning for all architectures.
Hi Jaroslav,
Yes, this is exactly what we were looking for - sorry I didn't even notice that in the bugzilla until you pointed it out.
Thanks!
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://access.redhat.com/errata/RHBA-2020:1883