Skip to content

Latest commit

 

History

History
31 lines (26 loc) · 1.87 KB

galera_questions_answers.md

File metadata and controls

31 lines (26 loc) · 1.87 KB
Raw

A few questions, with answers, on Galera

  • Is it possible, when creating an instance of a MySQL DB, to pass MySQL config params in the -c '{"binlog_row_image": "FULL"}' string?

    • ANS: We don’t support a feature like parameter groups or the configuring of binlog_row_image right now.

  • Is it possible, for any of the supported topologies, to do a zero-downtime upgrade (say, upgrade the tile, then each of the instances)? The Galera topology seems most likely to support this, but I’m not sure it’s possible. Maybe it is, and that’s just the default way it works.

    • ANS: I suppose it depends how tightly we define zero downtime. Galera should get close, but we have a few lingering bosh-dns integration issues which probably mean up to several seconds of downtime. Application connections will be dropped during an upgrade regardless and need to reconnect (potentially interrupting in-flight transactions).

  • It appears that the Galera option yields a specific number of nodes, and the traffic is round-robin distributed across these, with synchronous replication among them. Am I right?

    • ANS: Traffic is round-robin across the cluster using bosh-dns, but directed to proxies that send all queries to a single database node. This avoids failures due to cluster transaction certification conflicts which can significantly degrade cluster performance.

  • Again, referring to Galera, is it possible to add additional nodes, to scale the workload across more nodes?

    • ANS: A galera cluster can be scaled somewhat -- but it is not a write-scaling technology. This could be useful for scaling reads for specific workloads, but an app would need to be designed to be tolerant of stale reads or enable Galera specific options to do synchronous reads. We currently do not support that use case in the MySQL v2 product.