SpitfireLoad Testing & APMWhat's new

Infrastructure: IBM Cloud (Beta)

The Tower reads IBM Cloud Monitoring with a read-only service ID's API key: VPC virtual servers and load balancers, IKS and OpenShift clusters, Databases for PostgreSQL / MySQL / Redis / MongoDB, Event Streams and Cloud Object Storage. Nothing is installed in IBM Cloud and nothing is changed. In the license each Monitoring instance counts as one cloud account. The other clouds: AWS, Azure, Google Cloud, vCenter.

What you need

  • The IBM Cloud CLI (ibmcloud) on your computer, signed in as an account administrator (ibmcloud login --sso).
  • An IBM Cloud Monitoring instance in the account, with platform metrics turned on for its region. The cluster panels fill only when the cluster is connected to that instance; Event Streams consumer lag needs the Enterprise plan.
  • Outbound HTTPS from the Tower to *.cloud.ibm.com.

Quick install

  1. In All integrations In Spitfire: /admin/integrations choose IBM Cloud Monitoring, copy the command and run it in a terminal. It asks for the service ID name and the Monitoring instance name.
bash
curl -fsSL "https://<spitfire>/api/v1/setup/ds/ibm.sh?t=<token>" -o /tmp/sf-ibm.sh && bash /tmp/sf-ibm.sh
  1. When you see OK: ..., you are done. The first metrics come within about 5 minutes.

When the script cannot reach Spitfire it prints the values to enter. The API key is not printed; it goes to ~/.spitfire-setup/ibm-<date>.txt, readable only by you. Paste the value into the Spitfire field, then delete the file with the rm -f … command the script prints.

Manual install

  1. Sign in and check the account.
bash
ibmcloud login --sso && ibmcloud target
  1. Find the Monitoring instance's GUID and region (the GUID and Location lines).
bash
ibmcloud resource service-instances --service-name sysdig-monitor --long
  1. Create the service ID.
bash
ibmcloud iam service-id-create spitfire-monitor --description "Spitfire read-only"
  1. Give it Viewer + Reader on the Monitoring instance (reads metrics). Put the GUID from step 2 in place of <guid>.
bash
ibmcloud iam service-policy-create spitfire-monitor --roles Viewer,Reader --service-name sysdig-monitor --service-instance <guid>
  1. Give it Viewer on the account and its resource groups (to list servers, clusters and services).
bash
ibmcloud iam service-policy-create spitfire-monitor --roles Viewer && ibmcloud iam service-policy-create spitfire-monitor --roles Viewer --resource-type resource-group
  1. Create its API key and show the apikey value.
bash
ibmcloud iam service-api-key-create spitfire-monitor-key spitfire-monitor --description "Spitfire" --file "$HOME/spitfire-monitor-key.json" && grep '"apikey"' "$HOME/spitfire-monitor-key.json"
  1. Add it in Spitfire → Infrastructure sources → IBM Cloud Monitoring.
text
Region: from step 2 (e.g. eu-de) · Instance GUID: from step 2 · API key: the apikey value above · Test → Save
  1. Delete the key file.
bash
rm "$HOME/spitfire-monitor-key.json"
  1. Check the roles.
bash
ibmcloud iam service-policies spitfire-monitor | grep -q sysdig-monitor && echo "OK: spitfire-monitor reads IBM Cloud Monitoring"

Fields

Field What to enter
Region The Monitoring instance's region: eu-de, us-south, jp-tok…
Instance GUID The Monitoring instance's GUID (step 2)
API key The service ID's API key (stored encrypted, never shown again)
VPC regions Regions whose virtual servers are listed; empty = the region above
Services Which services are read: vsi,k8s,postgresql,mysql,redis,mongodb,eventstreams,cos,lb
API address Empty for IBM Cloud; only for an API gateway in between
Interval 1, 5 (default), 10 or 15 minutes

What is read

Service Metrics
VPC virtual servers (vsi) CPU, memory, network in/out, disk read/write; running or not
IKS / OpenShift (k8s) Ready nodes, node CPU and memory, pod restarts; cluster state and workers
Databases (postgresql, mysql, redis, mongodb) Connections, CPU, memory, disk
Event Streams (eventstreams) Bytes in/out, partitions, offline partitions, consumer lag
Object Storage (cos) Used bytes, objects, requests, 5xx errors
VPC load balancers (lb) Throughput, connections, latency, back-end 5xx, unhealthy pool members

Finding thresholds: virtual server CPU 90%, database disk 85%. A stopped virtual server, a cluster not in the normal state, unhealthy load balancer members, back-end 5xx, Event Streams offline partitions and consumer lag also raise findings.

When the test fails

Message What to do
IAM refused the API key The key was deleted or mistyped: repeat step 6 and paste the new key
The service ID lacks a role Repeat steps 4 and 5
Instance not found Check the GUID and the region (step 2)
The region does not exist The region code is wrong (e.g. eu-de, not Frankfurt)

An account that cannot be reached is skipped for that read; the next read tries again. If no data arrives, check two things first: the region code (eu-de, not a city name) and platform metrics being on in that region.

Uninstalling

The setup creates only the spitfire-monitor service ID in IBM Cloud, its policies and its API keys. If you chose another name, use it.

  1. Delete the source in Spitfire: All integrations In Spitfire: /admin/integrations#observability → the bin on its row → Delete. Metrics already collected stay.
  2. Delete the service ID; IBM Cloud deletes its API keys and policies with it.
bash
ibmcloud iam service-id-delete spitfire-monitor -f
ibmcloud iam service-id spitfire-monitor >/dev/null 2>&1 || echo 'OK: spitfire-monitor removed'
  1. Delete the local files the script left.
bash
rm -f ~/.spitfire-setup/ibm.state ~/.spitfire-setup/ibm-*.txt && echo 'OK: local files removed'