StarFerry Upgrade
StarFerry is the third major upgrade to the LikeCoin chain
Last updated
StarFerry is the third major upgrade to the LikeCoin chain
Last updated
Upgrade cosmos-sdk to 0.45.6, ibc-go to 2.3.0
Introduce x/likenft module, and integrate x/nft module backported from cosmos-sdk 0.46.0-rc1
Testnet Proposal
W4 June 2022
Testnet Deploy
W1 July 2022
Mainnet Proposal
W1 July 2022
Mainnet Deploy
W2 July 2022
FoTan
v1.2.0
.liked/cosmovisor/genesis/bin/liked
LaiChiKok
v2.0.2
.liked/cosmovisor/upgrades/v2.0.0/bin/liked
StarFerry
v3.0.0
.liked/cosmovisor/upgrades/v3.0.0/bin/liked
The system resources required during the upgrade will be higher than normal operation for data migration tasks.
In particular, large amount of memory will be used, and data files will be duplicated for backup.
After the upgrade is applied, the day-to-day resources requirement is same as previously suggested in our Node Setup Guide.
Below recommendation is based on recent test runs of migration with mainnet data. Any machine with less than 16GB will likely result in OOM (out of memory) error. Also note that using swap will lengthen the upgrade duration considerably.
Recommended
8 CPU Cores
64GB RAM
500GB+ SSD
Minimum
4 CPU Cores
32GB RAM + 32GB Swap
500GB+ SSD
Using RAM lower than 16GB will likely result a OOM and corruption of data.
Operators who could not arrange a capable machine might opt to temporarily stop validating until the network has upgraded, then catch up to trusted nodes using state sync.
v3.0.0 is tagged on repo and binary builds are released on github via CI
Upgrade proposal is raised on-chain, which specifies the following parameters:
Upgrade Name: v3.0.0
Upgrade Height: (TBC)
Upgrade Info:
After the proposal is approved, operators should place the v3.0.0 binary at the Cosmovisor upgrades folder BEFORE the upgrade time. Alternatively, operators can configure cosmovisor to download binaries automatically.
Cosmovisor will automatically switch binaries at the scheduled block height. Operators will not be required to run migrations or restart chain manually. However, operators are recommended to be present during the upgrade time in case of any unforeseen issues.
Migrate to Cosmovisor
Operators can also test their setup and practice the migration flow on testnet before the actual upgrade.
Upgrading without Cosmovisor
Although not recommended, it is possible for operators to opt out of using Cosmovisor.
At the upgrade height, the old chain binary will halt with upgrade needed panic. Afterwards, operators shall backup their .liked folder, then start their node using the new version binary.
Prepare the next binary ahead of upgrade height
Once the proposal is approved on-chain, operators should either download or build the next binary version, and place it within the Cosmovisor upgrades folder.
The path to the new binary should be:
Alternatively, operators can enable the auto-download feature of Cosmovisor by setting environment variable DAEMON_ALLOW_DOWNLOAD_BINARIES to true, which is the default if our setup script or docker image was used. To be safe, we recommend you do prepare the binary manually.
Review node configuration
Due to Cosmos SDK changes, there is a requirement of configuring minimum gas price. In the past, the configuration is empty string (zero gas price) by default. After the upgrade, the node will refuse to start up if a gas price is not explicitly set. We recommend setting a non-zero value, which can mitigate spam transactions on the network.
We also urge all validators to take this chance to review their app.toml config, especially the state-sync snapshots feature. By contributing state-sync snapshots, the synchronization process of new nodes can be speeded up from days to minutes. This will be vital to the growth of our network, as the full chain data size is growing rapidly due to more features and applications being added to LikeCoin chain.
Below are items to be updated in app.toml:
Backups
Cosmovisor will backup the .liked/data folder to .liked/data-backup-<date> after the chain has halted at upgrade height and before the new binary has started.
Contingency Steps to Skip or Rollback Upgrades
Below are steps to skip the upgrade forcefully. This workflow can be used under the following circumstances:
A critical bug is found shortly before upgrade height, thus there is no time to cancel the upgrade on-chain via CancelSoftwareUpgrade proposal
The upgrade failed on mainnet and the community reached consensus to rollback the chain to the state before the upgrade height
To skip an upgrade on short notice, operators shall append --unsafe-skip-upgrades <upgrade height> to their cosmovisor launch argument.
Set launch argument for systemd
Stop the node by sudo systemctl stop liked
Edit /etc/systemd/system/liked.service
Locate line beginning with ExecStart=, append --unsafe-skip-upgrades <upgrade height> to the end
Save the file
(For rollback, restore .liked folder with additional steps in the next section)
Run sudo systemctl daemon-reload, then sudo systemctl restart liked
Restore .liked folder to rollback upgrade
This is in addition to adding the skip upgrade flag step above.
Point the current binary symlink to the previous version (v2.0.0 for this upgrade):
If the community decided to rollback after blocks were produced post-upgrade, we will need to restore the data folder from backup:
CI/CD Release
Once a release candidate is validated on devnet / testnet, please tag the commit with version v3.0.0. CI will test and build with the tagged commit, then release it on GitHub.
Please also push commits to release/v3.x and update branch names in node setup guides.
Submit Proposal
The proposer should submit a proposal with below command:
The proposal content, especially upgrade-info should be validated on testnet beforehand.
Refer to for steps to add swap space.
To support the automatic upgrade, operators should migrate to host their node with Cosmovisor beforehand, which is supported since v1.2.0. A setup guide with migration instructions is available .
Please also refer to the technical documentation on upgrade proposal usage .