What causes extended latency between blocks?

Some people have been pointing out that the timestamps for some consecutive blocks can be far more than 4 seconds apart. Blocks 23292486 and 23292487 appear to be about 21 seconds apart. Urtho mentioned in the discord that this apparently happens fairly often, with block times ranging from 6 to 22 seconds roughly every couple hundred blocks (Discord). All I’ve heard so far is that nodes can fail sometimes, requiring block selection and confirmation to be redone. This is my first post here, so if there’s anything I can do better, please let me know!

1 Like

It’s likely caused by clock differences between the block proposers. Clock synchronization is not enforced by the protocol. It just requires that the timestamp is within 25s of the previous block: specs/ledger.md at master · algorandfoundation/specs · GitHub

1 Like