3 ms·
Good point: instead of thinking "what should I do if the other computer failed", you go straight to the "I'll do this and that if you don't prove me that you're
by woliveirajr 11y ago
Good point: instead of thinking "what should I do if the other computer failed", you go straight to the "I'll do this and that if you don't prove me that you're doing good in 50ms"
- MichaelGG 11y agoIt gets more complicated with shared state, though. Think of a DB split brain. If the slave incorrectly takes over then stuff gets corrupted. So you need proper fencing, and that alone seems like it's non-trivial. Common solutions seem to add a single point of failure (iSCSI fencing) or rely on the network being robust enough (quorum).
- tkinom 11y agoSyncing complete config db can be very simple too. For me it was just one line of wget code to fetch data over and untar it. The process took < 1 seconds. After which, all the configure and mirrors to standby in C code. A LOT more code was written to test the config + HA switch over process. The test script was continually configure the system on the master side via snmp and SW force switch over every 30 seconds and run over night, over weekend, and forever to see if there was any unexpected configure errors and failures.