-
Notifications
You must be signed in to change notification settings - Fork 32
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Browse files
Browse the repository at this point in the history
* Create meeting notes sig-chain #24 Signed-off-by: George Appiah <[email protected]> * Update 024-2025-02-11.md added video recording link Signed-off-by: Drinkwater <[email protected]> --------- Signed-off-by: George Appiah <[email protected]> Signed-off-by: Drinkwater <[email protected]> Co-authored-by: Drinkwater <[email protected]>
- Loading branch information
1 parent
b4da7e9
commit 9243bcc
Showing
1 changed file
with
136 additions
and
0 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,136 @@ | ||
| # Akash Network - Chain Special Interest Group (SIG) - Meeting #24 | ||
|
|
||
| ## Agenda | ||
| - Updates on Akash blockchain and network upgrades | ||
| - Migration from Cosmos SDK 45 to 47 | ||
| - Performance issues and indexing optimizations | ||
| - Introduction of an intermediate network upgrade (v0.38.0 ) | ||
| - Updates on Testnet 6 and next steps | ||
|
|
||
| ## Meeting Details | ||
| - Date: Tuesday, February 11, 2025 | ||
| - Time: 08:00 AM PT (Pacific Time) | ||
| - [Recording](https://c3kekwiljt434lc5ivhynrbkhea5ktklaa5hq3ejkkmrtt6uzqwa.arweave.net/FtRFWQtM-b4sXUVPhsQqOQHVTUsAOnhsiVKZGc_UzCw) | ||
| - [Transcript](#transcript) | ||
|
|
||
| ## Participants | ||
| - Tyler Wright | ||
| - Artur Troian | ||
| - Scott Carruthers | ||
| - Alex Pedersen | ||
| - B S | ||
| - CriptoRuim | ||
| - Damir Simpovic | ||
| - Floreat Flourish | ||
| - Rodri R | ||
| - Zeke Ezagui | ||
|
|
||
| ## Meeting Notes | ||
|
|
||
|
|
||
| ### Introduction and Overview | ||
| - Tyler Wright opened the meeting, acknowledging ongoing testing efforts for the Cosmos SDK 47 migration. | ||
| - Thanked community members for their contributions to Testnet 5 and the initial phase of Testnet 6. | ||
| - Handed the discussion over to Artur for updates on network upgrades and testing. | ||
|
|
||
| ### Performance Issues and Indexing Optimizations | ||
| - Artur highlighted unexpected performance issues discovered during Testnet 6: | ||
| - Akash state size has grown significantly due to increased activity (console trial wallets, deployments, and transactions). | ||
| - Exported state from genesis now exceeds **15GB**, leading to inefficiencies in query performance. | ||
| - Current state indexing does not differentiate between open and closed orders, requiring full state scans to retrieve active items. | ||
| - Similar inefficiencies exist for leases and bids, making provider startup times slow. | ||
| - Filtering mechanisms actually **decrease** performance due to inefficient indexing structures. | ||
| - Additional issues were identified with console grant key queries, which rely on inefficient granter-based indexing. | ||
|
|
||
| ### Introduction of an Intermediate Network Upgrade (v0.38.0 ) | ||
| - Artur announced an **intermediate upgrade (v0.38.0 )** before the SDK 47 migration. | ||
| - This upgrade will implement **optimized state indexing**, allowing: | ||
| - State objects to be stored under prefixes to improve lookup times. | ||
| - Reverse indexes for active leases and bids to reduce scan times. | ||
| - Expected to **greatly improve performance** of order, lease, and bid queries. | ||
| - aep-61 document detailing these improvements has been published and will soon transition from **draft** to **final**. | ||
|
|
||
| ### Additional Performance Fixes and Dependencies | ||
| - Discovered an issue with large genesis files (>4GB) causing failures in **Tendermint’s Snappy compression plugin**. | ||
| - Fix was identified in newer versions of **CometBFT** and needed to be backported to maintain compatibility. | ||
| - Testing of these fixes delayed SDK 47 work but has now been completed. | ||
|
|
||
| ### Status of Cosmos SDK 47 Migration | ||
| - Due to the indexing changes, SDK 47 migration must wait until v0.38.0 is finalized and deployed. | ||
| - Remaining tasks for SDK 47: | ||
| - Integration of AkashJS with the Akash API for TypeScript clients. | ||
| - Final migration of the **provider codebase** to SDK 47. | ||
| - Estimated time for completion: **one additional week of development** once v0.38.0 is fully tested. | ||
|
|
||
| ### Next Steps | ||
| - Artur and the core engineering team will: | ||
| - Finalize **aep-61 indexing improvements** and push for network-wide adoption. | ||
| - Continue local testing of v0.38.0 before deploying to a sandbox environment. | ||
| - Prepare a **software upgrade proposal** for the Akash blockchain. | ||
| - Resume SDK 47 migration work after v0.38.0 is validated. | ||
|
|
||
| ### Community Involvement and Closing Remarks | ||
| - Tyler emphasized the importance of community feedback on **aep-61** before finalization. | ||
| - Encouraged validators and developers to **stay engaged in Discord** for future updates on testing and network upgrades. | ||
| - The recording and transcript of this meeting will be shared publicly, including in validator and SDK 47 migration discussion channels. | ||
| - Tyler thanked Artur and all contributors for their efforts and closed the meeting. | ||
|
|
||
| ## Action Items | ||
| - **Artur Troian:** Complete aep-61 indexing improvements and prepare v0.38.0 for deployment. | ||
| - **Tyler Wright:** Share meeting recording and ensure aep-61 feedback is collected. | ||
| - **Community Members:** Review aep-61 and provide feedback before finalization. | ||
| - **Core Engineering Team:** Resume SDK 47 migration work after successful v0.38.0 deployment. | ||
|
|
||
| ## **Transcript** | ||
|
|
||
| Tyler Wright: All right, welcome everybody to Sig Chain. is February, no idea what today's date is. it is February 11th, 2025. second week of the month. Again, the SIG chain is a monthly meeting that meets the second Tuesday of each month to go over all things related to the Kosh blockchain, network upgrades, anything like that. | ||
|
|
||
| Tyler Wright: We have some other meetings where we talk about any issues in the core codebase. You can go to six support for that. But again, specifically here, we usually have conversations with some validators and other members of the community about all things related to the class blockchain. I know that there's been some ongoing testing around the Cosmos SDK 47 migration. I know there's been a couple of items that the community has been testing as a part of test net 5. So, thank you to the community that helped out with one of testing as we get ready for the cashless SDK 47 migration. I know we got moved into phase two and we're doing some testing in test net 6 that was ongoing for a month and I think we've run into a couple of blockers that we're addressing. | ||
|
|
||
| Tyler Wright: so I'll hand it over to Archer to give us some updates on where we are with the Cosmos SDK 47 migration and any other potential updates that he might have related to the blockchain and network upgrades. | ||
|
|
||
| Artur Troian: Thanks. Hello everyone. Raichy long longstanding cosmos 47. so we have a little bit of backlog with that and we pretty much were ready to start doing the tests on the sandbox. However, I think two weeks ago, we did a bit of discovery related to the performance of the quiring of the blockchain. So, pretty much to summarize what has happened, the Akash state has grown up quite dramatically in the last few months. contributions to that are the console trial wallets and so on so forth. | ||
|
|
||
| Artur Troian: just to give you perspective right now exported state from state into the genesis file. So the new genesis file if you try to export all of the records is over like 15 gigabytes and just insane. Anyways we needed to do optimizations into the existing mainet indexing. So what happens is that if let's say I want to query bids or let's say active orders from the current state. The current state does not differentiate the orders by state. It's pretty much only single index by the order ID. | ||
|
|
||
| Artur Troian: And in order to find open orders on the network, it has to loop through the entire state. just to give you perspective, right now it's how I'm actually testing. So I think there's couple million orders in the stage. You can imagine how long it takes to move through all of them and filter out just the open. And That applies for the leases and with the bits and leases in particular it's a bit complicated because provider needs to know its leases at start up and it won't obviously know only active leases and once again it's exactly the same problem. It to blockchain RPC node has to loop through all of the leases on the network to find a active leases b for the specific provider. | ||
|
|
||
| Artur Troian: I want to ask there's a filtering but the problem with the filtering is it actually slows things down because once again there's only one index and the filtering pretty much makes things only worse. So there's another issue related to the Z quering performance. console needs to find the grants by grant key and how this store index is built only on the granter. So the granter is the sort of like a main key. | ||
|
|
||
| Artur Troian: we have been sort of this debating where about the Z and we were planning to put the fixed in upcoming XDK47 but then we discovered all those performance issues related to the quering with the orders deployments and entire market stores and bits leaders and orders and we decided it makes sense to make intermediate network upgrade. So we published AP61 describing the issue and the solution. Let me just share it real quick. | ||
|
|
||
|
|
||
| ### 00:05:00 | ||
|
|
||
| Artur Troian: And the solution is to introduce first of all item number one split state of the objects put the state of the object into the prefix. So it does not increase amount of records on the state but it uses the state of the object as the indexing key which allows us effectively to cut all unnecessary items that we trying to search through and then in certain cases for leaders and bits we are implementing the reverse index it's going to introduce some data duplication but | ||
|
|
||
| Artur Troian: We introduce those reverse indexes on only items that are actually necessary active leases and open and active bids. So we won't need it on the closed leases or closed bits for example. let me just share screen now. So everybody could just have a comments on this AP. I think we have finalized all of the indexes and all of the items here in the description and I will be measuring it in the next couple days. So if you have any comments please also if not we will move it from the draft status into the final and start preparing for the network upgrade. | ||
|
|
||
| Artur Troian: So gi given that right now we are testing extensively the upgrade of the state and that upgrade of the state has discover has surfaced one more issue and that issue is related it's not actually related to the main net but it's related to the testing. So what happens is the tender mean when it initializes the node start from the genesis it stores the genesis into the state for a number of reasons and the problem is when the genesis state grows over 4 GB in size the snappy plugin which does encoding and compacting from the level DB does not store chunks more than 4 GB and it was panicking | ||
|
|
||
| Artur Troian: So the newer versions of the comic BFT do have fix for that. as we see on tender mean 3427 we had to backport it and that porting needs to do some testing so on so forth. So that pushed a few items related to this upgrade and SDK47 behind. all of those issues have been resolved. So we have all the back ports to the common BFT as well as the Cosmos SDK and now pretty much we just testing network great locally. As soon as we confident that all of the tests are passing we're going to push prey releases to the network repo and start testing it against sandbox. As soon as we have all the things aligned, we push into the software upgrade proposal to the blockchain. | ||
|
|
||
| Artur Troian: does anybody have any questions related to that? moving on now to the SDK47. so as you could probably imagine the SDK47 was kind of blocked by this upcoming network upgrade for a couple of reasons. One of them is because due to the state changes the implementation of the SDK 47 upgrade now needs to be adopted and we don't want to adopt it until we confident that current intermediate upgrade is fully functioning. So as soon as we publish software upgrade proposal we start resumeuming work on SDK47. There are two items remaining in the SDK 47. | ||
|
|
||
| Artur Troian: Item number one is integrating Akash GS parts into the Akash API that's for TypeScript clients and the second part is finishing m migrating of the provider codebase into the new SDK because provider codebase depends on the blockchain obviously as well. we need probably a week to finalize it. So my hope is that once we start testing the current intermediate upgrade which is going to be called V038 again sandbox the team starts working on the resumes work on the SDK47. I think that pretty much wraps it up. Any more questions? A Russian ter. | ||
|
|
||
|
|
||
| ### 00:10:00 | ||
|
|
||
| Tyler Wright: Thank you very much, Archer, for the update. I'm going to make sure that this recording is made available along with the notes in the transcript as quickly as possible and I'm going to drop it in channels like validators so that other people that may not be able to join this meeting regularly because of timing issues can still get up these updates. but does anyone else have any other questions about anything that Archer talked about or anything blockchain or network upgrade related? Cool. I know that we just had someone that just joined us. | ||
|
|
||
| Tyler Wright: again for anybody that just joined we just got an update from the core engineering team around Cosmos SDK migration some of the blockers that came up there and there's going to be again an intermediate network upgrade prior to the bigger Cosmos SDK 47 migration network upgrade. The recording from this meeting will be made available shortly and I'll drop it in validators channel and other channels and maybe even in the Kas SDK 47 migration discussion just for reference point for people but again if anybody has any specific questions or anything that they want to talk about as it pertains to the Akos network blockchain or upcoming network upgrades feel free | ||
|
|
||
| Tyler Wright: to ask those questions If there's no questions, then again, I'm going to make this recording available very soon so that folks in the community that can't join this meeting or maybe joined a little bit late can get an update on all things related to the intermediate network in the Cosmos SDK47 migration network upgrade that'll be soon following. But I know there's a great deal of work that's been going on as pertains to core engineering is getting these continuing to improve the blockchain. As I mentioned Archer shared 861. | ||
|
|
||
| Tyler Wright: so if anybody has any comments or thoughts feel free to drop them in there before that is merged. Again thank you to all those that participate in net 5 in the beginning phase one of test net 6. Look out for announcements as thing as work continues to get done. Look out for announcements as it pertains to next steps for the intermediate network upgrade. but again much appreciated everyone's time today. Look out for more announcements coming soon in a number of channels and I'll make sure this recording is shared publicly. but thank you again Archer and thank you everyone else. Have a great rest of your day. Bye-bye. | ||
|
|
||
| Artur Troian: I survive. | ||
|
|
||
| Zeke Ezagui: Thanks. | ||
|
|
||
| Rodri R: It's my lake. | ||
|
|
||
|
|
||
| ### Meeting ended after 00:14:09 👋 | ||
|
|