NO-JIRA: feat: add ssh-node target for quick SSH access to cluster nodes #55
There are no files selected for viewing
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,92 @@ | ||
| #!/bin/bash | ||
|
|
||
| set -euo pipefail | ||
|
Contributor
There was a problem hiding this comment. This is going to make a few lines of helpful errors unreachable, I'll point them out below |
||
|
|
||
| SCRIPT_DIR=$(dirname "$0") | ||
| DEPLOY_DIR="$(cd "${SCRIPT_DIR}/../.." && pwd)" | ||
| INVENTORY_FILE="${DEPLOY_DIR}/openshift-clusters/inventory.ini" | ||
|
|
||
| usage() { | ||
| echo "Usage: $0 <node> [command]" | ||
| echo "" | ||
| echo "SSH into a cluster node via the hypervisor jump host." | ||
| echo "If a command is provided, execute it and return." | ||
| echo "" | ||
| echo "Node can be specified as:" | ||
| echo " master-0, master_0, node0, 0 -> first master node" | ||
| echo " master-1, master_1, node1, 1 -> second master node" | ||
| echo "" | ||
| echo "Examples:" | ||
| echo " $0 master-0 # Interactive SSH session" | ||
| echo " $0 0 uptime # Run 'uptime' on master-0" | ||
| echo " $0 1 'pcs status' # Run 'pcs status' on master-1" | ||
| exit 1 | ||
| } | ||
|
|
||
| if [[ $# -lt 1 ]]; then | ||
| usage | ||
| fi | ||
|
|
||
| NODE_ARG="$1" | ||
| shift | ||
|
|
||
| # Check if inventory file exists | ||
| if [[ ! -f "${INVENTORY_FILE}" ]]; then | ||
| echo "Error: Inventory file not found at ${INVENTORY_FILE}" | ||
| echo "Run 'make inventory' first." | ||
| exit 1 | ||
| fi | ||
|
|
||
| # Normalize node argument to inventory name | ||
| case "${NODE_ARG}" in | ||
| master-0|master_0|node0|0) | ||
| NODE_PATTERN="master_0" | ||
| ;; | ||
| master-1|master_1|node1|1) | ||
| NODE_PATTERN="master_1" | ||
| ;; | ||
| *) | ||
| echo "Error: Unknown node '${NODE_ARG}'" | ||
| usage | ||
| ;; | ||
| esac | ||
|
|
||
| # Extract node IP from inventory | ||
| NODE_IP=$(grep -E "master_0|master_1" "${INVENTORY_FILE}" | grep "${NODE_PATTERN}" | grep -oP "ansible_host='\\K[^']+") | ||
|
Contributor
There was a problem hiding this comment. I don't think this'll happen, but maybe we need to parse for the full word (master_0 or master_1, as this will match master10, master100, and also things like ostest_master_10) Suggestion might be changing grep -E to grep -W to parse only full words |
||
|
|
||
| if [[ -z "${NODE_IP}" ]]; then | ||
|
Contributor
There was a problem hiding this comment. Might be unreachable: if line 55 grep doesn't find anything, it will have exit code 1 and script should exit. Adding "|| true" to the command on 55 should be enough, we're checking for content of grep output anyway |
||
| echo "Error: Could not find ${NODE_PATTERN} in inventory" | ||
| echo "Make sure the cluster is deployed and inventory is updated." | ||
| exit 1 | ||
| fi | ||
|
|
||
| # Extract hypervisor info from inventory | ||
| HYPERVISOR=$(grep -E "^[^#]*@" "${INVENTORY_FILE}" | head -1 | awk '{print $1}') | ||
|
Contributor
There was a problem hiding this comment. This is great to parse the ProxyJump, but it might also hit the @ in "ansible_ssh_common_args" if the inventory were ordered differently, so although it works with the current inventory structure, it's a bit fragile. |
||
|
|
||
| if [[ -z "${HYPERVISOR}" ]]; then | ||
|
Contributor
There was a problem hiding this comment. Might be unreachable: if line 64 grep doesn't find anything, it will have exit code 1 and script should exit |
||
| echo "Error: Could not find hypervisor in inventory" | ||
| exit 1 | ||
| fi | ||
|
|
||
| # Common SSH options to avoid known_hosts issues after redeploys | ||
| SSH_OPTS=( | ||
| -o StrictHostKeyChecking=no | ||
| -o UserKnownHostsFile=/dev/null | ||
| -o LogLevel=ERROR | ||
|
Contributor
There was a problem hiding this comment. I would move this to WARN to make debugging easier |
||
| ) | ||
|
|
||
| # ProxyJump with same options for the jump host | ||
| PROXY_CMD="ssh ${SSH_OPTS[*]} -W %h:%p ${HYPERVISOR}" | ||
|
|
||
| if [[ $# -gt 0 ]]; then | ||
| # Run command and return | ||
| ssh "${SSH_OPTS[@]}" \ | ||
| -o "ProxyCommand=${PROXY_CMD}" \ | ||
| "core@${NODE_IP}" "$@" | ||
| else | ||
| # Interactive session | ||
| echo "Connecting to ${NODE_PATTERN} (${NODE_IP}) via ${HYPERVISOR}..." | ||
| ssh "${SSH_OPTS[@]}" \ | ||
| -o "ProxyCommand=${PROXY_CMD}" \ | ||
| "core@${NODE_IP}" | ||
| fi | ||
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
the $(cmd) might be expanded as an "empty" argument to make if it's not provided. To allow for no command safely, might need something like this:
@./openshift-clusters/scripts/ssh-node.sh $(node) $(if $(cmd),$(cmd))