1.2.6 Release (Important info!)

[sargonas]

We’ve released version 1.2.6, which is to address a bug-fix with nodes. While this is technically a bug-fix version, we want to stress the importance of this update being applied to your own nodes, and encourage you to update to it at the earliest convenience.

Last week a bug was detected in the nodes that caused them to consume excessive amounts of memory when the queue of incoming compacted-blocks from Bluebox Timelords grew faster than the node was able to process them. As this queue increased, it could eventually grow to consume all available memory, resulting in node and system instability. (If you’re unfamiliar with Bluebox Timelords and what they do, I cover that down below!) While this bug has been around since the launch of mainnet, this fully came to light last week because Chia Network rolled out 100+ Bluebox Timelords temporarily to both bring down the ever growing blockchain DB size, as well as to put into production the new Bluebox Timelord code we made recent improvements to.

Unfortunately, this caused nodes all over to receive a large quantity of compressed blocks, and the work of swapping the increased volume of compressed blocks into the DB for the uncompressed versions caused some nodes to be unable to keep up, and experience higher-than-acceptable memory usage. As a result, we brought those Blueboxes offline and developed this update. To address this problem, this update contains better handling logic on the node side to manage that incoming data in a way that allows the node to maintain stability, which should prevent this from happening again.

As the chain is continuing to grow and the bloat still needs to be addressed, Chia still intends to ramp back up to this same number of temporary Bluebox Timelords to compress things down for you and streamline the db size, but this time we will do a gradual increase of them over several weeks starting next week. As a result, anyone who has not updated to version 1.2.6 will potentially experience issues with their node if it’s queuing is unable to handle the rate of updates… while we would rather you did not experience this pain at all, we feel there is a greater user pain in the size of the current DB and the time it takes to sync fresh, and we are motivated to deploy this temporary cluster to bring it down for you.

Further context and full details can be found here: https://www.chia.net/2021/09/10/important-version-1.2.6-information.en.html