Dance moves disputed? Get off the Dance Floor! #1229
Conversation
| ok; | ||
| CloseState = blockchain_ledger_state_channel_v2:close_state(LedgerSC), | ||
| case {CloseState, SCDisputePrevention} of | ||
| {dispute, true} -> {error, already_disputed}; |
There was a problem hiding this comment.
all below is shifted over to allow for this clause.
| @@ -2387,7 +2577,7 @@ add_and_gossip_fake_blocks(NumFakeBlocks, ConsensusMembers, Node, Swarm, Chain, | |||
|
|
|||
| setup_meck_txn_forwarding(Node, From) -> | |||
| ok = ct_rpc:call(Node, meck_test_util, forward_submit_txn, [From]), | |||
| ok = ct_rpc:call(Node, blockchain_txn_mgr, submit, [fake_txn, fun(_, _) -> ok end]), | |||
| ok = ct_rpc:call(Node, blockchain_txn_mgr, submit, [fake_txn, fun(_) -> ok end]), | |||
There was a problem hiding this comment.
I think this might brake some tests as we are now using blockchain_txn_mgr:submit/2 in state channel workers
There was a problem hiding this comment.
Verifying we're thinking of the same thing
https://github.com/helium/blockchain-core/blob/mj/sc-close-disputes/src/state_channel/blockchain_state_channels_worker.erl#L175-L184
The place that was using the callback/2 version was https://github.com/helium/blockchain-core/blob/mj/sc-close-disputes/test/meck_test_util.erl
It was sending the Txn along with the ok result.
|
This seems to do the trick. |
|
|
||
| ok = blockchain_ct_utils:wait_until_height(RouterNode, 16), | ||
|
|
||
| %% REVIEW: How can I assert something here about the rewards? |
There was a problem hiding this comment.
I didn't see here what I was expecting. Any insight would be helpful.
There was a problem hiding this comment.
You can ask for blockchain_txn_rewards_v2:rewards_metadata and you should get back all of the gory details about rewards. It's what ETL uses to populate DB info about reward payouts.
|
|
||
| ok = blockchain_ct_utils:wait_until_height(RouterNode, 16), | ||
|
|
||
| %% REVIEW: How can I assert something here about the rewards? |
There was a problem hiding this comment.
You can ask for blockchain_txn_rewards_v2:rewards_metadata and you should get back all of the gory details about rewards. It's what ETL uses to populate DB info about reward payouts.
|
|
||
| %% REVIEW: How can I assert something here about the rewards? | ||
| %% Nothing has been disputed yet. | ||
| {ok, Rewards1} = ct_rpc:call(RouterNode, blockchain_txn_rewards_v1, calculate_rewards, [5, 16, RouterChain]), |
There was a problem hiding this comment.
Use v2 here b/c that's what we use in production
|
so many commits here, should probably be squashed before merge |
A flood of disputes coming in from all over the place can cause enough trouble reconciling how to pay out rewards and txns that things start to go haywire. - consistent arity for txn callback - xor filters don't like to be empty - set default sc_handler if you're not running tests with a group - helper for adding traces - add sc_dispute test - break out of valid check early if the ledger state channel is already disputed - add sc_dispute_prevention chaing var --boolean() - add sc_dispute_prevention eunit test for rewards - make sure dispute_prevention is active before rejecting close disputes - Zero out when dispute prevention is active and sc is dipusted, otherwise carry on as usual - better searchable name for the test - more explicit naming for internal test halpers - add sc_dispute_prevention chain var to test suite - Some notes and questions about this test - grab sc_dispute_prevention from passed in Vars - don't use assertMatch, allows empty maps to match any map - more descriptive variable - use rewards v2 Update src/transactions/v2/blockchain_txn_rewards_v2.erl Co-authored-by: Andrew Thompson <andrew@hijacked.us> sc_dispute_prevention -> sc_dispute_strategy_version - Allows for evolution of handling state channel close disputes - Add sc_dispute_strategy_version to rewards_vars() map - get back metadata for rewards - allow versions higher than 1 without code updates in these paths
michaeldjeffrey
force-pushed
the
mj/sc-close-disputes
branch
from
16de626
to
5873aa4
Compare
Vagabond
force-pushed
the
mj/sc-close-disputes
branch
from
98f1b09
to
a093ef3
Compare
is_valid check has to be done within the grace period, rewards can't be calculated until after the grace period. having the rewards check before will cause the test to always fail on the is_valid check with a cannot_expire result.
The added state channel test
sc_dispute_prevention_test, if there are dial failures, the test will fail. It will complain about not enough summaries. I do not know how to increase the reliability at this time.