DFINITY Foundation

Internet Computer Association

After recent issues [1] with increased manifest computation times in subnet `jtdsg`, leading to newly joining and/or fallen behind replicas no longer being able to successfully catch up to the most recent subnet state, a subnet management proposal to slow down the block rate of the subnet to 0.5 blocks per second was submitted and accepted [2]. Slowing down the block rate of the subnet resolved the immediate issue and allowed replicas to catch up again, but clearly a solution to the problem that does not require reducing the block rate of the subnet was needed. In [1] necessary optimizations to achieve this were described. Meanwhile most of them are implemented and included in the most recently elected replica version [3], and a proposal to upgrade subnet `jtdsg` to this replica version was accepted [4]. This brings manifest computation times to a range that allows to increase the block rate again. Concretely, the recent updates have reduced the manifest computation time for this subnet from approximately 5 minutes (when [2] slowed down the subnet) to 15 seconds now. 

This proposal restores the block rate of subnet `jtdsg` to 1 block per second, reverting the change of [2].

[1] https://internetcomputer.org/blog/recent-issues-on-subnet-jtdsg-and-upcoming-improvements

[2] https://dashboard.internetcomputer.org/proposal/29035

[3] https://dashboard.internetcomputer.org/proposal/38541

[4] https://dashboard.internetcomputer.org/proposal/39595


Update configuration of subnet: jtdsg