Created
Feb 28, 2022 3:00 PM
Last Edited Time
Feb 28, 2022 3:00 PM
Type
Weekly Sync
Created By
Participants
SnapDeals testing goals:
- Validate the SnapDeals flow in Lotus and Actors - We can start the Calibration run from next week
- Determine testing scenarios
- Determine testing timeline
Discussion around the topic
- We need to cut RC Lotus RC on Calibration since that means we need to enter code freeze.
- @jennijuju when do we actually plan to cut Lotus RC and enter code freeze?
We will do it for @Aayush Rajasekaran and @Dragan Zurzin when we really want to start
Important dates:
- @January 3, 2022 we start running RCs on Calibration net since we need to have things running for 5, 6 weeks, due to Venus and other implementations
- @December 20, 2021 - we will start testing with @Reiers, do a fast discovery and see what we get, re-calibrate and run it again. @ToBeRemoved will run the testing effort, @jennijuju will provide feedback.
Testing Scenarios we want to go over until the end of the Year(2021) :
- Happy paths we want to check - @jennijuju can add more:
Pack multiple deals into 1 sector and we make sure everything checks out
The sector is moved into the next state if the deal is set to expire
What happens to miners if WindowPost - Snapdeals lock to GPU needs to be tested + checked with @Łukasz Magiera why do we lock it in the first place?
I have a CC snap deals sector that is marked with 180 days, we have deals coming - we have 1 deal into the sector. I extend the sector by 100 days...and a deal thats end the epoch is around 220 days, What happens?
how are we doing with QAP? are we making sure SP gets the maximum / or close to maximum benefit?
@ToBeRemoved - I avoided this by giving SP power to adjust it. We are trying to automate it.
We need to review it since this is a good point.
The testing paths we want to validate from @ToBeRemoved :
Most basically check that your miner can do a snap deals upgrade and not see major problems with the storage market or power table
- No power loss after upgrade (window post correct)
- Block elections work after upgrade (winning post correct)
- No market faults/data loss after upgrade (storage market interactions correct)
One particular usability situation I'm worried about is the case when your miner is trying to upgrade sectors but all of the deals incident to the miner are longer than the remaining lifetime of those sectors. I would like to see if head-of-line blocking clogs up your deal making in this situation.
- Non-happy path - miner operation errors and it would be good to see how robust upgrading is at handling them. You will know better but I'm imagining worker failures, miner out of funds, deals get dropped are examples of failures people run into. In these cases, I'd like to get things failing gracefully. In particular the sector should not be stuck in a bad state and it should not miss window posts and correctly service winning post challenges.
March 2025
Today
Sun
Mon
Tue
Wed
Thu
Fri
Sat
23
24
25
26
27
28
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
1
2
3
4
5
Action items from the call
We will do it for @Aayush Rajasekaran and @Dragan Zurzin when we really want to start
We need to document the SP providers question around snapdeals - @Reiers
Actors bugs need to be fixed, next week we will hope to close this - @Aayush Rajasekaran @ToBeRemoved is on it
Migration needs work, we should test it for calibration - @Aayush Rajasekaran already worked on this - should be done during calibration - @January 20, 2022
@jennijuju will create a FIP - around this