Description of problem ====================== By default, gdeploy uses rhs-high-throughput tuned profile on both rhel 6 and rhel 7 based gluster storage servers. According to our docs, this is wrong approach: > Earlier versions of Red Hat Gluster Storage on Red Hat Enterprise Linux 6 > recommended tuned profiles rhs-high-throughput and rhs-virtualization. These > profiles are still available on Red Hat Enterprise Linux 6. However, > switching to the new profiles is recommended. List of new recommended tuned profiles (from table 11.1. Recommended Profiles for Different Workloads): * Large-file, sequential I/O workloads: rhgs-sequential-io * Small-file workloads: rhgs-random-io * Random I/O workloads: rhgs-random-io See section 11.2. Brick Configuration of Admin Guide: https://access.redhat.com/documentation/en-US/Red_Hat_Storage/3.1/html/Administration_Guide/Brick_Configuration.html Version-Release number of selected component (if applicable) ============================================================ gdeploy-1.0-12.el6rhs.noarch How reproducible ================ 100 % Steps to Reproduce ================== 1. Use gdeploy to setup Trusted Storage Pool without specifying tuned profile in gluster.conf file 2. Check tuned profile activated on gluster storage server machines using `tuned-adm active` command. Try this with both rhel 6 and rhel 7 storage server machines. Actual results ============== i) On rhel 6 storage servers, rhs-high-throughput profile was activated: ~~~ # tuned-adm active Current active profile: rhs-high-throughput Service tuned: enabled, running Service ktune: enabled, running ~~~ This is wrong default choice, because rhgs-sequential-io is now recommended. ii) On rhel 7, no profile is activated. Gdeploy fails on task called "Performance tune for general storage" with this message: ~~~ stderr: Requested profile 'rhs-high-throughput' doesn't exist. ~~~ As storage documentation points out, this is expected as the older profile is not available on rhel7 machines. Unfortunately this outcome is even worse, because admin would be confused and try to find problem in gluster.conf while the problem is in wrong default configuration. Expected results ================ On both rhel6 and rhel7, gdeploy uses rhgs-sequential-io as default tuned profile. Additional info =============== https://access.redhat.com/solutions/2023633 https://bugzilla.redhat.com/show_bug.cgi?id=1260597
This behavior has been changed in the next release version. Now, by default gdeploy does not assume a profile anymore. If not specified, tuned-adm will not be called at all. User should specify explicitly which profile is to be set, if at all needed.
See BZ 1249979, RHGSC is updating list of tuned profiles as well.
Fixed in release 1.1.
Tested for rhgs-sequential-io and rhgs-random-io. Verified with gdeploy-2.0-5.el7rhgs.noarch
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/RHEA-2016:1250