From a89332ab51049f08da6c8e901b062f66e70c982c Mon Sep 17 00:00:00 2001 From: "github-actions[bot]" <41898282+github-actions[bot]@users.noreply.github.com> Date: Fri, 30 Aug 2024 14:28:13 -0500 Subject: [PATCH] CASMCMS-9045 - IMS clarify configuring remote build nodes. (#5337) * CASMCMS-9045 - IMS clarify configuring remote build nodes. (cherry picked from commit a92aaad6febdc93880198fa0cc1f2562ef21870d) * Update based on PR feedback. * Update operations/image_management/Configure_a_Remote_Build_Node.md Signed-off-by: Nathan Rockershousen --------- Signed-off-by: Nathan Rockershousen Co-authored-by: David Laine Co-authored-by: Nathan Rockershousen --- .../Configure_a_Remote_Build_Node.md | 23 +++++++++++++++++++ .../Troubleshoot_Remote_Build_Node.md | 2 +- 2 files changed, 24 insertions(+), 1 deletion(-) diff --git a/operations/image_management/Configure_a_Remote_Build_Node.md b/operations/image_management/Configure_a_Remote_Build_Node.md index af4dcbcbc580..302d338ac5b7 100644 --- a/operations/image_management/Configure_a_Remote_Build_Node.md +++ b/operations/image_management/Configure_a_Remote_Build_Node.md @@ -61,6 +61,29 @@ There are only two requirements for using a compute node as a remote build node: * Have Podman installed and configured * Allow IMS access via SSH key +Each system is unique so the node chosen to run remote jobs needs to be manually configured based on the +requirements and capabilities of the system and that particular node. It is not possible to list all the +configuration steps to optimize a particular node. The following directions are only intended to make a +minimally functional remote node. Much better performance may be possible with custom configuration by a +system administrator who is familiar with the configuration and capabilities of the system. + +Some things to consider are: + +1. Volume mounts. + + There needs to be adequate space on the node to handle the images that are being created or customized + on the remote node. The faster the volume access, the faster the image customization. Ways to improve + performance beyond the mounts described below can include: + + * If there are local disk drives on the node, use them. + * If there is a lot of RAM available, create volumes from memory. + * If lustre is available, use lustre mounts. + +1. Network access. + + The package repositories that are needed for image creation and customization are located in Nexus + within the Kubernetes system. The remote node must be able to access Nexus in order to function. + ### Use an existing compute node This will add processes to the node being used as a remote build node. The system administrator diff --git a/operations/image_management/Troubleshoot_Remote_Build_Node.md b/operations/image_management/Troubleshoot_Remote_Build_Node.md index 5ad5c2d872b8..64fd3d094a00 100644 --- a/operations/image_management/Troubleshoot_Remote_Build_Node.md +++ b/operations/image_management/Troubleshoot_Remote_Build_Node.md @@ -61,7 +61,7 @@ are times that problems crop up with running these remote jobs. 1. (`ncn-mw#`) Check the IMS service logs for creation of the job. ```bash - cray -n services logs cray-ims- | grep -A 15 POST + kubectl -n services logs cray-ims- | grep -A 15 POST ``` When the job is created there is information in the log about matching remote build nodes - look