My AIT-2 experience & the cost

[RmznDndr]

I've divided this post into 3 parts: i. Server Specs ii. My AIT-2 Experience iii. GCP & Cost

i. Let's start with server specs.

Provider: GCP Location: Sydney OS: Ubuntu 20.04 ROM: 250 GB SSD RAM: 8 GB at start, upgraded to 12 GB later vCPU: 4 vCPU (E2 machines of GCP) at start, upgraded to 6 vCPU later Network: Premium tier

I've decided to use GCP for my validator node because of obvious reasons. I thought it was wise to do so especially considering Google's service quality and the juicy $300 trial credit. Overall experience with GCP was fine, i guess.

---

ii. My AIT-2 Experience

I could easily start my journey with the information provided by e-mail.. My node was online couple hours after i received the airdrop mail. And started to sync right after turning on. It was smooth. I've followed all the update request of the team.

testnet_41e4066d6500df4c026407b1b461d1581f13c360 commit caused my node to slow proposal & vote rate. I've slept as one of the top 30 participants and woke up as one of the below 100 on the leaderboard. My node still had 100% liveliness & 100% participation rate. I thought that it could be caused of my node specs being low compared to others.

testnet_95dae1f1221571c4cbca8f9a9c7b6e0182603c9c & testnet_a0290ec4b34d916ab51e874ae210cf6175a4fa72 commits were released on the same day. I applied them both. I was still at 100ish places on the leaderboard if i remember correctly.

The day i've updated my node to testnet_f5d8013b0a1851da8e078394d83130d3adaf7670 commit, my participation rate dropped below %90. It kept dropping after the update. It went down around %80. I got couple node panic and constant restarting issues, manually stopped the docker & restarted it but didn't help. I've removed the docker volume and started from stracth. It started to recover participation rate slowly after.

Around 8:50 UTC on 21th July, my node stopped working suddenly again. I was on discord trying to help others and noticed that my node was down. Created this issue: https://github.com/aptos-labs/aptos-core/issues/2125 Solution was the same again. Removing the docker volumes and starting from stracth. I've added the following code for faster syncronization and i believe it also helped me a lot.

state_sync:
  state_sync_driver:
    bootstrapping_mode: DownloadLatestAccountStates # Tells state sync to skip all historic transactions
  data_streaming_service:
    max_concurrent_requests: 3 # Enables data pre-fetching so syncing isn't super slow

---

iii. GCP & Cost

My original thought was using $300 trial credit of GCP & paying the extra costs as i go. I wasn't expecting anything higher than $350-400 including trial credit. And yet I got billed for ~4900 TRY extra (approx. $300). So my total cost was $600 ($300 trial credit + $300 bill). My node used all $300 trial credit on first 3 days.

I'm currently unable to extract the details of these billings but i'll add them in the future if i can. I believe that the main reason for high bill was extreme bandwidth usage. Because the main plan was $95/month. The only explanation is network usage.

Starting days were covered by trial credit and not included in this photo. Also the today's cost is not shown yet.

This is the current payment section.

[sherry-x]

thank you so much for all the info!!