Technical writing on Kubernetes, Ceph, PostgreSQL high availability, Slurm, GPU scheduling and infrastructure operations — including the parts where the conventional answer is wrong.https://quorate.esgmedia.com.hk/en-ushttps://quorate.esgmedia.com.hk/blog/backups-you-have-never-restored/https://quorate.esgmedia.com.hk/blog/backups-you-have-never-restored/The backup job going green every night is the most reassuring and least informative signal in infrastructure. Here is a method for turning it into a number you can quote, and the failure modes that hide behind a successful job.Sat, 10 Oct 2026 00:00:00 GMTBackup & DROperationsRiskQuoratehttps://quorate.esgmedia.com.hk/blog/kubernetes-dra-gpu-clusters/https://quorate.esgmedia.com.hk/blog/kubernetes-dra-gpu-clusters/Dynamic Resource Allocation replaces the device plugin model with something closer to how GPUs really work — but the operational questions it answers are narrower than the marketing suggests. Here is what changes, and what does not.Thu, 08 Oct 2026 00:00:00 GMTKubernetesHPC & AIGPUQuoratehttps://quorate.esgmedia.com.hk/blog/postgresql-ha-patroni-repmgr-pgpool/https://quorate.esgmedia.com.hk/blog/postgresql-ha-patroni-repmgr-pgpool/These three tools are frequently compared as if they solve the same problem. They do not. One manages cluster state, one manages replication, and one is a router that people mistake for a failover mechanism.Fri, 02 Oct 2026 00:00:00 GMTDatabasesPostgreSQLHigh availabilityQuoratehttps://quorate.esgmedia.com.hk/blog/ceph-crush-map-mistakes/https://quorate.esgmedia.com.hk/blog/ceph-crush-map-mistakes/Durability in Ceph comes from the CRUSH map, not from the replication number. A three-replica pool whose copies land on three OSDs in the same host is a one-copy pool with extra ceremony. These are the configurations we find most often.Fri, 25 Sep 2026 00:00:00 GMTDistributed storageCephFailure domainsQuoratehttps://quorate.esgmedia.com.hk/blog/why-your-ha-cluster-has-never-failed-over/https://quorate.esgmedia.com.hk/blog/why-your-ha-cluster-has-never-failed-over/Fencing is not a feature of high availability — it is the precondition for it. A cluster that cannot isolate a failed peer will, on a split, run both sides and write to the same data. Here is how to find out whether yours can.Fri, 18 Sep 2026 00:00:00 GMTHigh availabilityPacemakerFailure domainsQuorate