Thoughts on (not) automating the setup of our first cloud server: Sometimes it’s ok to not automate (or start with automation). There’s a lot of overhead that comes along with running in AWS (for eg) the way they want you to. If you don’t need it and can avoid allowing you to work on your actual problem … Well that just sounds great to me! Btw Automating for cloud vendor x does not make you portable to cloud vendor y. There are differences.
Green is the color of complacency: Well isn’t this an interesting read. Just because alarms aren’t going off, or you’re not on fire, doesn’t mean there isn’t something interesting happening right now or that everything is ok.
The Rot-Com Bubble: Investors in tech are aiming for growth over finding products people want or use. Sustainable, useful businesses are likened to being in “hell” if they don’t exhibit orders of magnitude growth / return on investment. Has a toxic feeling to it …
Investigating Mysterious Kafka Broker I/O When Using Confluent Tiered Storage: Nice sluething story for strange observed kafka storage tiering behaviour. Honeycomb must be processing a ludicrous amount of data …

A 10 years later reminiscence by oxide computer people for a big production event :)

HEAD-2006: sdc-oneachnode should not default to all nodes: Postmortem action after a incident where all physical machines in a datacenter were rebooted at the same time by mistake due to a tool with unexpected, error punishing default behaviour.
HEAD-2007: sdc-oneachnode should require command in a single argument: Another about command line parsing.
HEAD-2008: sdc-oneachnode should validate node list before execution: And 1 more for validation.