In my own opinion, as long as there are no explicit known issues, such as CVE issues, we could keep up with whatever the latest version of RT kernel CentOS provides. It means I vote #2.c. Of course, WR folks might have the best insight whether there are any potential issues on such a combined case that controller nodes run with standard kernel while a slight different RT kernel runs on the compute node. Regards, Yong From: "Lin, Shuicheng" <shuicheng.lin@intel.com> Date: Monday, 14 January 2019 at 10:28 AM To: "Somerville, Jim" <Jim.Somerville@windriver.com>, "Little, Scott" <Scott.Little@windriver.com>, "Wold, Saul" <saul.wold@intel.com>, "Rowsell, Brent" <Brent.Rowsell@windriver.com>, "Hu, Yong" <yong.hu@intel.com>, "Xie, Cindy" <cindy.xie@intel.com> Cc: "starlingx-discuss@lists.starlingx.io" <starlingx-discuss@lists.starlingx.io> Subject: RT kernel upgrade in CentOS 7.6 Hi all, For CentOS 7.6, the latest source rpm package we have for std and rt kernel are: Std kernel: kernel-3.10.0-957.1.3.el7.src.rpm Rt kernel: kernel-rt-3.10.0-957.rt56.910.el7.src.rpm You could find std kernel is 1 minor version ahead of rt kernel. And there is some discussion about it in patch [0]. There are some question I want to ask, and need your suggestion. Thanks in advance. 1st question: Is it acceptable to have minor different version for std and rt kernel? If not, why? 2nd question: If we decide to upgrade rt kernel to the same version as std. There are several way to do it: 1. We generate the src rpm based on code in GIT manually. And save the rpm package to CENGN server. So mirror downloader script could get the new src rpm as previous. 2. We switch to git code instead of src rpm package. We Add the rt kernel git to our manifest file, and download rt kernel code when do “repo sync”. 3. Keep current RT kernel first, and upgrade to new rt kernel when new src rpm is available. I don’t know why it is not available yet. :) [0]: https://review.openstack.org/625773 Best Regards Shuicheng