https://getfelix.dev/blog/ Notes from building Felix: how it works, why it is built the way it is, and what testing it has turned up. en https://getfelix.dev/blog/felix-0-5-to-0-6.html https://getfelix.dev/blog/felix-0-5-to-0-6.html Sat, 24 Oct 2026 00:00:00 +0000 What we found, fixed and added between Felix 0.5.0 and 0.6.0: the bugs that shared one shape, the ways we found them, the features that apps asked for and why, and what building it taught us. https://getfelix.dev/blog/even-overall-lopsided.html https://getfelix.dev/blog/even-overall-lopsided.html Fri, 23 Oct 2026 00:00:00 +0000 Felix placement kept every broker's leader count near the cluster-wide share, and a single stream could still land 1/1/10 with all its writes on one broker; this is how placement now spreads each stream's leaders too. https://getfelix.dev/blog/fast-without-losing-messages.html https://getfelix.dev/blog/fast-without-losing-messages.html Fri, 23 Oct 2026 00:00:00 +0000 How Felix answers a fsynced publish in a third of a millisecond, writes at the disk's full rate under load, and slows publishers down instead of dropping their messages. https://getfelix.dev/blog/before-you-loosen-the-timeout.html https://getfelix.dev/blog/before-you-loosen-the-timeout.html Sat, 17 Oct 2026 00:00:00 +0000 Intermittent failures in Felix's cluster tests kept turning out to be real bugs in the broker, so before loosening a timeout I now run the test against a commit from before the suspect change; three cases show what that finds. https://getfelix.dev/blog/acknowledged-then-forgotten.html https://getfelix.dev/blog/acknowledged-then-forgotten.html Sat, 17 Oct 2026 00:00:00 +0000 Under Felix's default periodic fsync, a crashed broker could come back missing records it had acknowledged while the cluster still counted it as caught up; this is how review and the TLA+ model found it on the follower and leader sides, and what the fix costs. https://getfelix.dev/blog/catching-a-one-percent-regression.html https://getfelix.dev/blog/catching-a-one-percent-regression.html Sat, 17 Oct 2026 00:00:00 +0000 Felix's rule is that a release must not be measurably slower than the one before it; this is how same-runner A/B runs, a Welch's t-test and throwaway bisect PRs in GitHub Actions find a 1-2% loss, and why a single benchmark run can't. https://getfelix.dev/blog/the-follower-that-forgot.html https://getfelix.dev/blog/the-follower-that-forgot.html Sat, 17 Oct 2026 00:00:00 +0000 On an idle shard nothing ships, so a leader never learned that a follower had lost its disk; this is how a storage identity per data directory fixed that, and why it changes only when a directory may hold less than its peers were told. https://getfelix.dev/blog/checking-nothing.html https://getfelix.dev/blog/checking-nothing.html Fri, 16 Oct 2026 00:00:00 +0000 Felix checks 107 TLA+ configurations in CI; this is about three times a passing check meant nothing, and the twin configurations, reachability checks and spec pairing we use to catch it. https://getfelix.dev/blog/one-slow-subscriber.html https://getfelix.dev/blog/one-slow-subscriber.html Fri, 16 Oct 2026 00:00:00 +0000 How Felix keeps one slow subscriber from holding up publishers or other readers, and how a 2% single-publisher loss came from where the feeder gave back a subscriber's byte budget. https://getfelix.dev/blog/the-data-came-back.html https://getfelix.dev/blog/the-data-came-back.html Thu, 15 Oct 2026 00:00:00 +0000 Deleting a Felix stream or cache and creating one with the same name could bring back the old one's data; this is how incarnations fixed it, and how we kept control-plane members on different releases from storing different answers during the rollout. https://getfelix.dev/blog/point-your-kafka-client-at-felix.html https://getfelix.dev/blog/point-your-kafka-client-at-felix.html Mon, 28 Sep 2026 00:00:00 +0000 Felix brokers speak enough of the Kafka wire protocol for Kafka producers and partition-assigning consumers to use durable Felix streams; this is how a Kafka request becomes a Felix operation, how idempotent sequences survive a failover, and exactly where compatibility stops. https://getfelix.dev/blog/ten-parcels-one-envelope.html https://getfelix.dev/blog/ten-parcels-one-envelope.html Sun, 27 Sep 2026 00:00:00 +0000 Raising Felix's QUIC MTU on Linux stalled delivery for good, because UDP GSO turns ten packets into one datagram the kernel refuses and quinn never falls back; this is how we found it, why the cap is 6,550, and why 4,096 is the default. https://getfelix.dev/blog/introducing-felix.html https://getfelix.dev/blog/introducing-felix.html Sat, 26 Sep 2026 00:00:00 +0000 What Felix is, how a publish travels from a client to disk to every subscriber, why it is built that way, and how to run it on your laptop in five minutes.