DEV Community: nikania

Updating parachain with migrations

nikania — Tue, 30 May 2023 11:44:20 +0000

How to look and add migrations while upgrading parachain to new version via runtime upgrade.

We use example of simple parachain node template with tag polkadot-v0.9.32 and upgrade it to polkadot-v0.9.40.
To test this upgrade we need local testnet. How to create simple testnet short guide here.

Build `polkadot-v0.9.32` parachain-template version

First, we add some pallets that changed between these two versions to parachain-template:
pallet-alliance and pallet_assets.
Then, we build and place parachain in /bin folder

cargo build --release
mkdir bin
cp ./target/release/parachain-template-node \
    ./bin/parachain-template-node-v0.9.32

Prepare for runtime upgrade

(official guide here)
a) change versions polkadot-v0.9.32 to polkadot-v0.9.40 in Cargo.toml
b) fix compile errors🤯
a1)-b1) it's faster to use polkadot-v0.9.40 parachain-template version instead upgrading in this example😎
c) change spec_version

d) add migrations

this was the most unclear to me (do I need to apply them? where to find migrations? how to apply them?), but I'll show you how to add substrate migrations (not your own):

look through the releases in Polkadot repo and search for migrations description, here I found migrations in v0.9.34 release, v0.9.38 release, v0.9.40 release
download the source code and look for migrations in runtime/polkadot/src/lib.rs in struct frame_executive::Executive, migrations can be summed in type Migrations or just put there, here I found the following migrations:

// in v0.9.34
pub type Migrations = (
    //9320
    // "Bound uses of call" <https://github.com/paritytech/polkadot/pull/5729>
    pallet_preimage::migration::v1::Migration<Runtime>,
    pallet_scheduler::migration::v3::MigrateToV4<Runtime>,
    pallet_democracy::migrations::v1::Migration<Runtime>,
    pallet_multisig::migrations::v1::MigrateToV1<Runtime>,
    // "Properly migrate weights to v2" <https://github.com/paritytech/polkadot/pull/6091>
    parachains_configuration::migration::v3::MigrateToV3<Runtime>,
    //9330
    pallet_election_provider_multi_phase::migrations::v1::MigrateToV1<Runtime>,
    pallet_fast_unstake::migrations::v1::MigrateToV1<Runtime>,
    //9340
    pallet_balances::migration::MigrateToTrackInactive<Runtime, xcm_config::CheckAccount>,
    crowdloan::migration::MigrateToTrackInactive<Runtime>,
);

// in v.0.9.38
pub type Migrations = (
    // "Use 2D weights in XCM v3" <https://github.com/paritytech/polkadot/pull/6134>
    pallet_xcm::migration::v1::MigrateToV1<Runtime>,
    parachains_ump::migration::v1::MigrateToV1<Runtime>,
    // Remove stale entries in the set id -> session index storage map (after
    // this release they will be properly pruned after the bonding duration has
    // elapsed)
    pallet_grandpa::migrations::CleanupSetIdSessionMap<Runtime>,
);

// in v0.9.40
pub type Migrations = (
    pallet_nomination_pools::migration::v4::MigrateToV4<
        Runtime,
        NominationPoolsMigrationV4OldPallet,
    >,
);

also look for migrations in Cumulus repo, here I found v9360 v9330 v9380 v9400
migrations found in Cumulus:

// v9400
pub type Migrations = (pallet_nfts::migration::v1::MigrateToV1<Runtime>,);
// v9380
pub type Migrations = (pallet_contracts::Migration<Runtime>,);
// v9360
type Migrations = (
    pallet_alliance::migration::Migration<Runtime>,
    pallet_balances::migration::MigrateToTrackInactive<Runtime, xcm_config::CheckingAccount>,
);
// v9330
pub type Executive = frame_executive::Executive<
    Runtime,
    Block,
    frame_system::ChainContext<Runtime>,
    Runtime,
    AllPalletsWithSystem,
    pallet_assets::migration::v1::MigrateToV1<Runtime>,
>;

for all pallet that exist in your chain, add migrations, here I added (don't need to add balances: read instructions in pallet_balances::migration::MigrateToTrackInactive):

pub type Migrations = (
    pallet_xcm::migration::v1::MigrateToV1<Runtime>,
    pallet_alliance::migration::Migration<Runtime>,
    pallet_assets::migration::v1::MigrateToV1<Runtime>,
// added myself just to look in logs
    TestOnRuntimeUpgrade,
);

pub struct TestOnRuntimeUpgrade;

impl OnRuntimeUpgrade for TestOnRuntimeUpgrade {
    fn on_runtime_upgrade() -> frame_support::weights::Weight {
        frame_support::log::info!("ಠ_ಠ 🤞🤞🤞🤞🤞🤞🤞🤞🤞🤞 migrations applied");
        Weight::from_ref_time(1)
    }
}

e) build cargo b -r, save wasm

Perform upgrade and check everything is ok

run old-versioned parachain in local testnet (can use quick list)

cargo run ./bin/parachain-template-node-v0.9.32

upgrade runtime (official guide)

Migrations of pallet_xcm, pallet_alliance and pallet_assets are needed only if there is something in the storage of these pallets. Therefore, in this example migrations are unnecessary. I did this example only to show how to find the migrations.

Any comments, corrections and suggestions are appreciated!💬

Quick list - starting local parachain testnet

nikania — Fri, 21 Apr 2023 15:24:51 +0000

Quick CLI reference to start local testnet with 2 relay chain nodes and 1 parachain node. Based on detailed official tutorial

To start relay chain, firstly build chain spec:

./target/release/polkadot \ 
build-spec --chain rococo-local --raw > rococo_local_raw.json

in the resulting file set "bootNodes": [] (and unique "id": "qwerty" - optional, only if there are other testnets in your local network)

Start Relay Alice node:

./target/release/polkadot \
--alice \
--validator \
--base-path /tmp/relay/alice \
--chain rococo_local_raw.json \
--port 30333 \
--ws-port 9944

Start Relay Bob node:

./target/release/polkadot \
--bob \
--validator \
--base-path /tmp/relay-bob \
--chain  rococo_local_raw.json \
--port 30334 \
--ws-port 9945

In case of restart, purge previous chain state:

rm -rf /tmp/relay/alice
rm -rf /tmp/relay-bob

To start parachain

build spec (again, set unique "id" if needed) and wasm and genesis:

./target/release/parachain-collator \ 
build-spec --disable-default-bootnode > rococo-local-parachain-plain.json
./target/release/parachain-collator \
build-spec --chain rococo-local-parachain-plain.json --raw \
--disable-default-bootnode > rococo-local-parachain-2000-raw.json

./target/release/parachain-collator \
export-genesis-wasm --chain rococo-local-parachain-2000-raw.json > para-2000-wasm
./target/release/parachain-collator \
export-genesis-state --chain rococo-local-parachain-2000-raw.json > para-2000-genesis

Start parachain

./target/release/parachain-collator \
--alice \
--collator \
--force-authoring \
--chain rococo-local-parachain-2000-raw.json \
--base-path /tmp/parachain/sora-parachain-alice \
--port 40333 \
--ws-port 8844 \
--log xcm=trace \
-- \
--execution wasm \
--chain .._path_to_relay_chain_spec_/rococo_local_raw.json \
--port 30344 \
--ws-port 9977

Submit sudo on relay chain:

Using log in Substrate

nikania — Wed, 10 Aug 2022 10:57:00 +0000

How to log in substrate pallets and how to see logs when running node (can be useful to see existing logs in polkadot when debugging).

Import log crate in Cargo.toml of pallets/template

[dependencies]
log = "0.4"

Add some logging to extrinsic do_something in pallet-template (gibberish in my case)

log::trace!(target: "lalala", "do_something is working!");
log::trace!(target: "ononono", "do_something is working!");
log::trace!(target: "test::werr", "do_something is working!");
...
log::trace!(target: "lalala", "do_something is done!");
log::trace!(target: "ononono", "do_something is done!");
log::trace!(target: "test::werr", "do_something is done!");

Build node, and run with log filtering (filters only target: "lalala" trace logs):

./target/release/node-template --dev --tmp --log lalala=trace

when execute extrinsics, trace log is in output:

different filtering (notice that target: "test::werr" can be filtered like that test=trace)

./target/release/node-template --dev --tmp --log lalala=trace ononono=trace test=trace

--different output

For example to trace xcm messages in parachains/relay chain filter can be used: --log xcm=trace

List of common coding errors using Substrate

nikania — Mon, 21 Mar 2022 15:51:39 +0000

I have started to study Substrate recently, and encountered a lot of simple but not-so-easy to solve errors. So I decided to make a list of them, for future use and for others)

error[E0432]: unresolved import pallet
This is compile error caused by errors in
pub mod pallet { <fix errors here> }
error: failed to download parity-db v0.2.2
can be resolved by:

cargo update -p parity-db

many errors while building substrate-node-template often can be resolved with

cargo update -p <package causing problem>

error

.iter().map(|x| x.amount).sum();
      the trait `Sum<<T as pallet_*::Config>::Balance>` is not implemented for `&<T as pallet_*::Config>::Balance`

can be resolved by using

.fold(T::Balance::default(), |acc, x| acc + x);

instead of .sum() OR if u use custom asset pallet and can make changes in it, adding trait Sum<Self::Balance> to Balance let u use .sum()

/// The units in which we record balances.  
type Balance: Member + Parameter + AtLeast32BitUnsigned + Default + Copy + Sum<Self::Balance>;

error 1002 (in polkadot.js.org/apps/) 1002: Verification Error: Runtime error: Execution failed: ApiError("Could not convert parameter tx between node and runtime: Could not decode Call::MyPallet.0:\n\tCould not decode Call::create.3:\n\t\tNot enough data to fill buffer\n")

This error appears because of wrong types.json - wrong type here: Call::create.3 - 3rd argument with start from 0 (0,1,2,3)

error