Monitoring in motion

Challenges in monitoring Kubernetes, containers and dynamic infrastructure.

Transcript

1. Monitoring In Motion Challenges in monitoring kubernetes, containers, and dynamic infrastructure. CNCF Paris Meetup 16 Nov, 2017 Haïssam Kaj haissam@datadog.com
2. Haïssam Kaj • haissam@datadog.com • @ha_kaj • Team lead, container ecosystem @ Datadog About me
3. • SaaS based infrastructure and app monitoring • Open Source Agent • Time series data (metrics and events) and traces • Processing trillions of data points per day • Intelligent Alerting • We’re hiring! (www.datadoghq.com/careers/) Datadog Overview
4. Operating Systems, Cloud Providers, Containers, Orchestrators, Datastores, Caches, Queues and more... Monitor Everything
5. Outline 1. Intro: The Importance of Monitoring 2. The Challenge: Monitoring Dynamic Infrastructure 3. Finding the Signal: How do we know what to monitor? 4. Wrapping up: Applying this to a Go app on Kubernetes
8. Collecting data is cheap;
 not having it when you need it can be expensive
9. Instrument all the things!
10. Source: http://bit.ly/1SvvbuP
11. Source: http://bit.ly/1RQRsXW
12. Operational Complexity Increases with.. • Number of things to measure
 • Velocity of change
13. https://www.datadoghq.com/docker-adoption/
16. How much we measure? 1 instance • 10 metrics from cloud providers 1 operating system (e.g., Linux) • 100 metrics 50~ metrics per application
18. How much we measure? 1 instance • 10 metrics from cloud providers 1 operating system (e.g., Linux) • 100 metrics 50~ metrics per application
 N containers • 150*N metrics
19. Operational Complexity 100 instances 700 containers
20. Operational Complexity: Scale 160 metrics per host 1120 metrics per host Assuming 7 containers per host
21. Operational Complexity: Scale 100 instances 112 000 metrics Assuming 7 containers per host
22. How much we measure? 1 instance • 10 metrics from cloud providers 1 operating system (e.g., Linux) • 100 metrics 50~ metrics per application
 N containers • 150*N metrics Metrics Overload!
23. Operational Complexity Increases with.. • Number of things to measure
 • Velocity of change
24. Source: Datadog
25. Enter the Orchestrators
26. Operational Complexity Increases with.. • Number of things to measure
 • Velocity of change
27. Open Questions • Where is my container running? • What is the capacity of my cluster? • What port is my app running on? • What’s the total throughput of my app? • What’s its response time per tag? (app, version, region) • What’s the distribution of 5xx error per container?
28. Monitoring 101
29. Demo setup - github.com/hkaj/demo-app
30. Deployment Demo setup - github.com/hkaj/demo-app StatefulSet pod DaemonSet Service Service pod pod Deployment Service pod
31. Finding Signal - Categorizing Your Metrics
35. Recurse until you find root cause
36. What to demand from our monitoring tooling?
37. Host Centric
38. Service Centric
42. Query Based Monitoring “What’s the average throughput of application:nginx per version ?” “Alert me when one of my pod from deployment:foo is not behaving like the others?” “Show me rate of HTTP 500 responses from nginx” “… across all data centers” “… running my app version 2….”
43. Side Car agent
44. Getting at the metrics…
46. How do we get at the upper layers?
47. Resource Metrics Utilization: • CPU (user + system) • memory • i/o • network traffic Saturation • throttling • swap Error • Network Errors 
 (receive vs transmit)
48. Docker & Kubernetes Events • Starting / Stopping Containers • Scaling Events for Underlying Instances • Deploying a new container build
49. Containers
50. STATS Command # Usage: docker stats CONTAINER [CONTAINER...] $ docker stats $CONTAINER_ID CONTAINER CPU % MEM USAGE/LIMIT MEM % NET I/O BLOCK I/O ecb37227ac84 0.12% 71.53 MiB/490 MiB 14.60% 900.2 MB/275.5 MB 266.8 MB/872.7 MB
51. Docker API • Detailed streaming metrics as JSON HTTP socket
 $ curl -v --unix-socket /var/run/docker.sock http://localhost/containers/ 28d7a95f468e/stats

52. Pseudo-files • Provide visibility into container metrics via the file system. • Generally under: 
 /sys/fs/cgroup/<resource>/docker/$CONTAINER_ID/

53. Pseudo-files: CPU Metrics $ cat /sys/fs/cgroup/cpuacct/docker/$CONTAINER_ID/cpuacct.stat > user 2451 # time spent running processes since boot > system 966 # time spent executing system calls since boot $ cat /sys/fs/cgroup/cpu/docker/$CONTAINER_ID/cpu.stat > nr_periods 565 # Number of enforcement intervals that have elapsed > nr_throttled 559 # Number of times the group has been throttled > throttled_time 12119585961 # Total time that members of the group were throttled (12.12 seconds) Pseudo-files: CPU Throttling
54. Kubernetes
55. metrics - cluster • Generates metrics about the state of Kubernetes objects (nodes, pods, services, jobs…) • Network, disk stats • Leader status • Work metrics (consensus proposals, wal sync…) apiserver(s) • status check • resource metrics
56. metrics - node kubelet /healthz health check /pods metadata /api/v1.3/machine/ (cAdvisor) node resources
57. LAST SEEN NAME KIND REASON SOURCE MESSAGE 22m dd-agent-2pml8.14f45a4ece3aeca4 Pod Killing kubelet, gke-haissam-dl13 Killing container with id dd-agent:Need to kill Pod 21m dd-agent-482vl.14f45a5618aea4c0 Pod SuccessfulMountVolume kubelet, gke-haissam-wnvn MountVolume.SetUp succeeded for volume "cgroups" 21m dd-agent-482vl.14f45a5632a1e86d Pod Pulling kubelet, gke-haissam-wnvn pulling image "datadog/docker-dd-agent:latest" 21m dd-agent-482vl.14f45a5649590c91 Pod Created kubelet, gke-haissam-wnvn Created container 21m dd-agent-482vl.14f45a5650fb2dfd Pod Started kubelet, gke-haissam-wnvn Started container 22m dd-agent.14f45a4ea0acb0c0 DaemonSet SuccessfulDelete daemon-set Deleted pod: dd-agent-2pml8 19m nginx-deployment Deployment ScalingReplicaSet deployment-controller Scaled down replica set nginx-569477d6d8 to 0 events
59. Aren’t we still missing a layer?
60. Auto Discovery Docker API Kubernetes Monitoring Agent Container A O A O Containers List & Metadata Additional Metadata (Tags, etc) Config Backends Integration Configurations Host Level Metrics
62. Custom Metrics • Instrument custom applications
 • PUSH • STATSD • DogStatsD • PULL • Go Expvar, Prometheus, JMX, …
63. Monitoring Questions • Where is a given container running? • What is the overall capacity of my cluster? • What port(s) are my applications running on? • What’s the total throughput of my application? • What’s its response time per tag? (app, version, data center) • What’s the distribution of 5xx error per container? What about by data center?
64. Resources Monitoring 101: Alerting 
 https://www.datadoghq.com/blog/monitoring-101-alerting/ Monitoring 101: Collecting the Right Data https://www.datadoghq.com/blog/monitoring-101-collecting-data/ Monitoring 101: Investigating performance issues https://www.datadoghq.com/blog/monitoring-101-investigation/
 The Power of Tagged Metrics https://www.datadoghq.com/blog/the-docker-monitoring-problem/ 8 surprising facts about Docker Adoption https://www.datadoghq.com/docker-adoption/ Control groups, part 4: On accounting https://lwn.net/Articles/606004/
65. Q&A more questions? haissam@datadog.com @ha_kaj