brooks
@brooksdotsol
engineer @anza_xyz
setup switching between internal sched and pack. it's so easy, and i'm pumped for how quick deploying changes is going to become. not packing in this example, just logging status messages from scheduler & received tpu packets. switch to, from, and back to all within 10 seconds.
IBRL
This was fun thanks a lot @0xIchigo! Also made me realize sometimes when I ramble I sound like Trump? Bullish I guess?
Run an RPC node? Have secondary indexes enabled? Startup in v2.3.4 will be faster for you.
You can just ship things. Restart in *under* 3 minutes.
3 minutes to start and load ledger (this is going to go down a lot with the upcoming io-uring impl) 5 minutes to be back to tip 🔜 agave 2.3
Agave v2.3 is almost here. Here are the most important updates so you can monitor the situation: - Faster epoch transitions - AccountsDB optimizations to reduce I/O overhead - Greedy scheduler enabled by default - RPC API updates to simulate data usage - New TPU client
Alpenglow is cool, but it won't ship for a long time ❌ Wrong. Here's a 10 node Alpenglow cluster pulling 20k TPS
Poof! As of epoch 804, the epoch accounts hash is now gone. Enjoy some spare perf
Just merged a PR that'll make startup times on v2.3 one to two minutes faster. IBRL.
I confirmed with the *real* Alex that this is a fake account. Please unfollow/report/block immediately.
sorry folks this is an urgent upgrade. we need to disable a vulnerable, but currently unreachable feature. details to follow. do it now (soon)
Operators upgrade to agave or jito version 2.2.16 or to frankendancer version 0.505.20216 asap 🚨 See mb-announcements in discord for more details
Operators upgrade to agave or jito version 2.2.16 or to frankendancer version 0.505.20216 asap 🚨 See mb-announcements in discord for more details