Acceptance DAS test + DAS fixes #179
Conversation
Nashatyrev
changed the title
Nashatyrev
changed the title
|
Relates to #164 |
| @@ -65,7 +65,7 @@ public SafeFuture<InternalValidationResult> onDataColumnSidecarGossip( | |||
| return validation.thenPeek( | |||
| res -> { | |||
| dasGossipLogger.onReceive(dataColumnSidecar, res); | |||
| if (res.isAccept()) { | |||
| if (res.isAccept() || res.isSaveForFuture()) { | |||
There was a problem hiding this comment.
when it's safe for future there are a lot of checks missing, it's not valid
There was a problem hiding this comment.
Oh, right! Then it complicates things a lot 🤔
Need to think more about it then
There was a problem hiding this comment.
Reverted this change and filed a separate issue: #180
| createTekuBeaconNode( | ||
| createConfigBuilder() | ||
| .withRealNetwork() | ||
| .withDasExtraCustodySubnetCount(128 - 4) |
There was a problem hiding this comment.
it will cap it on 128 anyway
i'd replace this to 128 for readability
There was a problem hiding this comment.
Does it look better this way: 291cf5e ?
| lateJoiningNode.start(); | ||
| lateJoiningNode.waitForGenesis(); | ||
| lateJoiningNode.waitUntilInSyncWith(primaryNode); | ||
| lateJoiningNode.waitForEpochAtOrAbove(6); |
There was a problem hiding this comment.
I have added this to check that sync is not lost after 2 epochs
|
|
||
| int epochSlots = primaryNode.getSpec().slotsPerEpoch(UInt64.ZERO); | ||
| int endSlot = 6 * epochSlots; | ||
| assertAllBlocksExistWithoutForks( |
There was a problem hiding this comment.
why do we need this if they are in sync?
There was a problem hiding this comment.
We are checking that no blocks missed and no forks happened. Else it would mean that something went not that right
There was a problem hiding this comment.
Oh, nah I'm tolerating empty slots now. Probably worth to rename this method.
But we definitely don't expect forks here as we have just one validator node
| @@ -30,6 +30,9 @@ public interface DataColumnSidecarRetriever { | |||
| */ | |||
| SafeFuture<DataColumnSidecar> retrieve(DataColumnSlotAndIdentifier columnId); | |||
|
|
|||
| /** Starts all the queued retrieve requests immediately */ | |||
| void flush(); | |||
There was a problem hiding this comment.
I'm not sure on this way to go. So we just call nextRound on this but it doesn't gurantee that the thing we submitted recently will be executed soon. It could be deep in the queue or whatever.
There was a problem hiding this comment.
Yeah, probably the javadoc could be more accurate? Or do you have any other approach in mind?
… for Custody to not download it again
…ration to accommodate shorter slot configs (like 2 sec per slot)
… to initiate the sampling without timeout implied by 1s delay in SimpleSidecarRetriever.netRound() loop
Nashatyrev
force-pushed
the
das-pr/electra-acceptance-test
branch
from
291cf5e to
607aebf
Compare
…rimarily for Custody to not download it again" This reverts commit 916c126.
PR Description
UPD: reverted and created separate issue: When gossiped column is SAVE_FOR_FUTURE, save it #180