DIY Fault-Tolerant Microcluster
summary
Describes how to build a fault-tolerant micro-cluster using minimal hardware: L2 topology for internal component links and L3 setup for external service availability (HSRP, corosync). Covers highly-available local block storage using DRBD in dual-primary mode with OCFS2 on top, running virtual machines in the micro-cluster, routine VM maintenance operations, solution limitations, and common problems with their fixes.
AI-written summary · about
chapters (28)
- 0:00Microclusters introduction
- 4:07Physical network architecture
- 5:51Cluster file system choice
- 7:13Disk array configuration
- 8:36Network interface setup
- 10:25Fault-tolerant block device
- 11:32DRBD sync setup
- 12:40GFS2 install and config
- 15:55Virtual machines creation
- 17:20VM updates and migration
- 18:46Pacemaker high availability
- 21:54IP failover setup
- 25:09Limitations and common issues
- 27:14Network connectivity issues
- 28:41Pacemaker agents improvement
- 30:08Pacemaker Corosync setup
- 31:12HA alternatives
- 33:21DRBD and DB modes
- 34:05Network file system benefits
- 34:48Cluster levels overview
- 36:14Corosync usage reduction
- 36:35VM scaling
- 37:40Node failure load balancing
- 38:02Cluster quorum issues
- 38:47Backups and snapshots
- 39:08GFS2 clustering problems
- 39:31Containers vs KVM
- 39:54Conclusion and Q&A
related talks
38mYour Own Little HA Cloud at the Edge (or at Home)
29mTowards Sustainable Multi-Cluster Management in Real-World Academic Infrastructure
21mSponsored Session: Fault-Tolerant Training: How We Build Rel...
13mEnhancing Cloud Scalability and Fault Tolerance
48mFault Tolerance on the Cheap Making Systems That Probably Won't Fall Over
20mRDMA at Cloud Scale: The OCI Experience
38mDeploying Database Clusters in the Cloud
16mFrom System TAYE7 to High Availability: What, Why, How?