Run bench on demo #1552
Run bench on demo #1552
Commits on Aug 28, 2024
-
-
Extract genDatasetConstantUTxO
To take faucet and client keys as arg, so that is reusable also for the current use-case.
-
So that is also reusable for the current use-case.
-
Extract findRunningCardanoNode'
To find a node runing on unknown network, like devnet. So that is also reusable for the current use-case.
-
Add network id as part of the options
Also removed starting node id as not needed, because Alice will always lead the demo bench.
-
-
Fix tx building for funding transaction
This solves the problem of: SubmitTxValidationError (TxValidationErrorInCardanoMode (ShelleyTxValidationError ShelleyBasedEraBabbage (ApplyTxError (UtxowFailure (UtxoFailure (AlonzoInBabbageUtxoPredFailure (FeeTooSmallUTxO (Coin 169153) (Coin 0)))) :| [])))) The problem was we were using a build-raw transaction api instead of one which autobalances it.
-
-
Extend generator of constant utxo property
To be valid for an arbitrary set of client keys, instead of just a single one.
-
Trying to work out why large numbers of clients fail
- Also note strange keys generated when allocating only small amounts of lovelace
-
-
-
Avoid passing network id to options and faucet keys on generators
As we forcing the bench to run on devnet for the moment.
-
Make bench-demo sim to run on top of previous run
For that we made the following changes: - avoid requesting for node's history upon connection - datasets transactions use sender same as recipient - finally we avoid waiting for nodes to be connected given that is assumed in this case
-
Replace hydra-keys options by hydra-client hosts instead
Given we connect to provided hydra-clients we don't need to check we observe a head is inicializing with the same parties as its assumed. Instead we take hydra-client hosts so that the solution scales in other settings.
-
-
-
-
-
-
-
-
- Build and run seperate nix steps
-
-
-
It seems to have the right priviledges to do the netem work
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Add persitency-dir options to hydra-node
Due to the watch_logs script isn't able to find it
-
-
-
Use local functions to seed the network
Instead of exposing private ones defined in where clauses.
-
Do not seed neetwork before runnin bench-demo
Because the script will be seeding the network. The reason for this its because the scripts does not rely on a genesis funding transaction.
-
Remove FIXME as its not reproduceable
It just simply does not break ¯\_(ツ)_/¯
-
Minor refactor on generateOneTransfer
So that is reusable for this use-case without braking others.
-
-
-
-
Refactor HydraClient to keep peer information
This is useful to triger node actions not depending on localhost, like in demo-bench where commit has to be done on running peers.
-
-
Generate each client dataset from a different client funding tx
Instead of usign a single one from the faucet current utxo.
-
-
-
-
-
-
Make funding tx optional for datasets
This is used to seed the network in other scenarios. This one assumes the cluster is pre-seeded.
-
-
-
-
-
-
-
-
Instead, let the scenario seed the network and provide funds to running clients. At the end, funds are returned back to the faucet.
-
-
-
add failure handling on tx processing while head is open
this allows us to close the head when we get a time exceeded while wait for all confirmations when processing a tx.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Commits on Aug 29, 2024
-
Remove hack to check every node sees head is initialized for demo
In this case the parties are unknown but we still need to fetch the headId from a head id observation.
-
Fix not seeding during scenario but before it gets executed
The reason is depends on before starting the hydra-cluster. That consumes the Faucet UTxO making the fundingTransaction from dataset invalid, thus making the scenario to fail.
-