Skip to content

Commit 2b9102b

Browse files
envestccenvestcc
committed
postmortem-2023-10-10
1 parent 806e826 commit 2b9102b

File tree

1 file changed

+107
-0
lines changed

1 file changed

+107
-0
lines changed

‎postmortem/2023-10-10.md‎

Lines changed: 107 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,107 @@
1+
---
2+
Postmortem ID: 6
3+
Title: Stake duration bonus not enabled for nft bucket
4+
Author: Chen Chen <chenchen@iotex.me>
5+
Created: 2023-10-10
6+
Updated: 2023-11-09
7+
---
8+
9+
10+
# Abstract
11+
12+
During the testing of the Staking as NFT(IIP-13) feature on Mainnet, it was found that the vote calculation of the NFT Bucket was incorrect. This resulted in a slight decrease in the calculation of delegate total votes and the distribution of bucket rewards. To fix this issue, a hard fork upgrade is required.
13+
14+
# Cause
15+
16+
The votes for the NFT Bucket only calculate the staked amount and does not consider the weighting of stake duration and stake-lock.
17+
18+
**Incorrect calculation**
19+
20+
$$votes=amount$$
21+
22+
**The correct**
23+
24+
$$votes=amount*(1+\log_{1.2}{(duration/86400)(1+m)})$$
25+
26+
$$m= \begin{cases} 1 & locked \\ 0 & unlocked \geq 1 \end{cases}$$
27+
28+
# Solution
29+
30+
Using the accurate formula for vote calculation post hard-fork height. State patching is unnecessary as NFT votes are calculated in real-time, instead of being stored in the database. Also ensure the consistency of the vote calculation formula across blockchain, rewarding service, and web/mobile.
31+
32+
We calculate the rewards for NFT staking from epoch 37021 to 39613 based on the new reward rules and provide users with a one-time supplemental payment for the difference. The details are shown below:
33+
34+
| Reward Address (io) | Reward Address (eth) | Rewards (IOTX) |
35+
| ----------------------------------------- | ------------------------------------------ | -------------- |
36+
| io1q77yjjytau2t0yrs656djzu6xe63umuj9afq6j | 0x07bc49488bef14b79070d534d90b9a36751e6f92 | 226.1097975 |
37+
| io19jg5h2r5m9qfpwswdat8jzacadk5cljl9j4m98 | 0x2c914ba874d94090ba0e6f56790bb8eb6d4c7e5f | 218.281543 |
38+
| io1cdv420ksuerpgkh6wkdete7yq02ulnsp6usfdw | 0xc359553ed0e646145afa759b95e7c403d5cfce01 | 149.3072101 |
39+
| io1hwugvvl2aumf9jk64vrm98l0vglr923ntq5rfz | 0xbbb88633eaef3692cadaab07b29fef623e32aa33 | 109.3416226 |
40+
| io1ar4uhyg8l576vp2jl5wzz9wdvg73jhlglwtjad | 0xe8ebcb9107fd3da60552fd1c2115cd623d195fe8 | 81.54376056 |
41+
| io1yvvlkfcrzlkewye0dc6uaqc4gnvn5wfqkq0ctn | 0x2319fb270317ed97132f6e35ce831544d93a3920 | 42.81638365 |
42+
| io1e2rfj7mh4ejawp9dy2ns035305xyahlwmhc0c3 | 0xca86997b77ae65d704ad22a707c6917d0c4edfee | 39.67336968 |
43+
| io1x0d7ju9xmvrcj8u2g6glcpk0vqavcs5968d7j9 | 0x33dbe970a6db07891f8a4691fc06cf603acc4285 | 38.88607628 |
44+
| io17axg9h32tnjyk65vculc8u9jkegzjl4maevj84 | 0xf74c82de2a5ce44b6a8cc73f83f0b2b650297ebb | 34.01664689 |
45+
| io19s6f55ux5kuamsyfm7gn93y55tt5yn3c8tm6kj | 0x2c349a5386a5b9ddc089df9132c494a2d7424e38 | 9.705679169 |
46+
| io1cew5hztn9c5y3d26yuuwdwdl0wxpm3l42a3eta | 0xc65d4b89732e2848b55a2738e6b9bf7b8c1dc7f5 | 8.143316348 |
47+
| io19un9p5jx2r5x85m4uh8ylmphj22ee0h3myxvl3 | 0x2f2650d24650e863d375e5ce4fec3792959cbef1 | 7.258715476 |
48+
| io1e933nrdazcv55dwac6mvrn472qcnfvtvx3uuk6 | 0xc963198dbd16194a35ddc6b6c1cebe503134b16c | 6.437383934 |
49+
| io1wkflcfuupysc00a73h8590377azht4wh3x6lu5 | 0x7593fc279c092187bfbe8dcf42be3ef74575d5d7 | 6.26276931 |
50+
| io1dcuhfumkk8uznqery04t5cq6edamensask7ef8 | 0x6e3974f376b1f829832323eaba601acb7bbcce1d | 6.181853975 |
51+
| io1v7ts37jua2py826er9t3hr9tru4nne6csck2zl | 0x679708fa5cea8243ab5919571b8cab1f2b39e758 | 6.021873814 |
52+
| io1pr774gjrn8ecaug749288mrr3077eh54qfej9e | 0x08fdeaa24399f38ef11ea95473ec638bfdecde95 | 5.704900111 |
53+
| io14cntt59fx92erdm0h4x7jjmlu0lucf2umelpq2 | 0xae26b5d0a9315591b76fbd4de94b7fe3ffcc255c | 5.298415282 |
54+
| io1w4jxyy9vrlwzz0ahp2p3n2227a3hxwuna54adp | 0x75646210ac1fdc213fb70a8319a94af763733b93 | 4.904248324 |
55+
| io1q226ssx0tsw2z4uthtzcd62alq9fmm32ym72sd | 0x0295a840cf5c1ca1578bbac586e95df80a9dee2a | 3.231342886 |
56+
| io1leaukdnk42lf56eeev3l8f06g8hd0tgmep8z96 | 0xfe7bcb3676aabe9a6b39cb23f3a5fa41eed7ad1b | 2.562878204 |
57+
| io1k8rmm9jfq0p802qaa426g6gm00nmlsdr5zfa4g | 0xb1c7bd964903c277a81ded55a4691b7be7bfc1a3 | 1.841719759 |
58+
| io1jslfpm04ady9ht7vm9jl5fy930wa5t8utpwd5d | 0x943e90edf5eb485bafccd965fa24858bddda2cfc | 1.712134574 |
59+
| io16w526wlurq338h3jxr8a5gw7s7pwepq2lk8l7h | 0xd3a8ad3bfc182313de3230cfda21de8782ec840a | 0.418652519 |
60+
| io10dmh3t9uthk7w5kuc823fn96ldwg0m8wnfwy2t | 0x7b7778acbc5dede752dcc1d514ccbafb5c87ecee | 0.164680286 |
61+
| io19snrtfm6742akugdqg84rc5mcntc5e3dkupaad | 0x2c2635a77af555db710d020f51e29bc4d78a662d | 0.150885108 |
62+
| io15d0f4yzxg6ka6r4x6prdt3mflkcahlaqx7gtf6 | 0xa35e9a904646addd0ea6d046d5c769fdb1dbffa0 | 0.012317156 |
63+
| io1ftu2zrzl9gd3fxuxg806dylsaa2qjtudsafse7 | 0x4af8a10c5f2a1b149b8641dfa693f0ef54092f8d | 0.007601529 |
64+
65+
# Impact
66+
67+
The main impacts consist of three aspects:
68+
- On-chain data: NFT Bucket rewards are calculated incorrectly.
69+
- Partnerships: Integration progress with LSD partnership could be delayed.
70+
- Teams: Hard fork exhausts the team's human resources.
71+
72+
# Activities
73+
74+
- 2023-09-28 08:20 : Inconsistency found between the displayed Votes on the web and the blockchain
75+
- 2023-09-28 08:20 : After a small discussion, it was decided to align the web and reward distribution with blockchain
76+
- 2023-09-29 08:20 : Internal testing of Staking as NFT feature initiated on the mainnet
77+
- 2023-10-09 05:11 : Feedback received by the group regarding NFT staking bonus votes being zero
78+
- 2023-10-09 16:11 : After a wide-range discussion, it was decided to fix the vote counting method on blockchain and synchronize the changes to the web and rewarding service
79+
- 2023-11-06 07:59:45: Redsea hard-fork is activated
80+
81+
# Why does this issue happen?
82+
83+
- Why use incorrect formula
84+
- The focus was mainly on the processing logic of the NFT bucket itself, such as stake, restake-related procedures. The correctness of the vote counting logic was somewhat neglected.
85+
- It was assumed that it was equal to the amount without raising this issue.
86+
- The correct formula was not found, and the programming pattern needs to be improved.
87+
- Business background is not well understood
88+
- Why not discover during testnet
89+
- During unit testing, there was a misconception about vote counting, and the test case was incorrectly written. A formal test case review process is missing
90+
- Missing comparison of votes data during system integration testing(between chain and analystics)
91+
- Why not discover during code review
92+
- The PR was too large, making code review more challenging
93+
- The code review was not meticulous enough.
94+
95+
# How can we avoid it in the future?
96+
97+
- Raise any doubts you may have
98+
- Keep PRs as small as possible
99+
- Enhance Testing
100+
- Review test cases more carefully
101+
- Prepare a test plan for major upgrades
102+
- Monitor data consistency among multiple systems
103+
- Think as the perspective of user
104+
105+
# Appendix
106+
107+
- [staking fix weighted votes of contract staking bucket #3936](https://github.com/iotexproject/iotex-core/pull/3936)

0 commit comments

Comments
 (0)