Full rework of the BlockFetch logic for bulk sync mode #1179
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
amesgen
force-pushed
the
blockfetch/milestone-1
branch
from
bab4a6a
to
670ae32
Compare
facundominguez
force-pushed
the
blockfetch/milestone-1
branch
2 times, most recently
from
da90985
to
b3434e8
Compare
amesgen
reviewed
Aug 6, 2024
ouroboros-consensus/src/ouroboros-consensus/Ouroboros/Consensus/Storage/ChainDB/Impl/Types.hs
Outdated
Show resolved
Hide resolved
Comment on lines
+600
to
+606
| -- REVIEW: What about all the threads that are waiting to write in the queue and | ||
| -- will write after the flush?! |
There was a problem hiding this comment.
Indeed, there could be short time after the queue is flushed, but before the ChainDB is closed (and hence nothing new can be added to the ChainDB). I think this already exists on main.
I raised this on the IOG Slack, but I don't think we need to do anything about it in this PR.
There was a problem hiding this comment.
In that case I'd say: let's add a note and remove the REVIEW item.
ouroboros-consensus/src/ouroboros-consensus/Ouroboros/Consensus/Storage/ChainDB/Impl/Types.hs
Outdated
Show resolved
Hide resolved
facundominguez
force-pushed
the
blockfetch/milestone-1
branch
3 times, most recently
from
512f637
to
3d55d7e
Compare
* Addition of ChainSyncClientHandleCollection, grace period, and starvation event in BlockFetch * Plug `rotateDynamo` into `BlockFetchConsensusInterface` * Removal of `bfcMaxConcurrencyBulkSync` * Changes in blockfetch decision tracing
Port of IntersectMBO/ouroboros-network#2721 Co-Authored-By: Thomas Winant <[email protected]>
* Move Genesis-specific BlockFetch config to GenesisConfig * Introduce GenesisConfigFlags for interaction with config files/CLI * Add missing instances for Genesis configuration
* Mention that the objector also gets demoted * Edit note on Interactions with the BlockFetch logic * Expand the comments motivating DynamoInitState and ObjectorInitState Co-authored-by: Nicolas “Niols” Jeannerod <[email protected]>
* Run more repetitions of LoE, LoP, CSJ, and gdd tests * Print timestamps for node restarts * Disable boring timeouts in the node restart test * Wait sufficiently long at the end of tests * Expect CandidateTooSparse in gdd tests * Add a notice about untracked delays in the node restart test * Set the GDD rate limit to 0 in the peer simulator * Have the peer simulator use the default grace period for chainsel starvations * Relax expectations of test blockFetch in the BulkSync case * Allow to run the decision logic once after the last tick in the blockfetch leashing attack * Shift point schedule times before giving the schedules to tests * Accomodate for separate decision loop intervals for fetch modes * Accomodate for timer added in blockFetchLogic * Switch peer simulator to `FetchModeBulkSync` * Allow parameterizing whether chainsel starvation is handled * Add some wiggle room for duplicate headers in CSJ tests * Disable chainsel starvation in CSJ test
amesgen
force-pushed
the
blockfetch/milestone-1
branch
from
16e5690
to
150ff7e
Compare
facundominguez
force-pushed
the
blockfetch/milestone-1
branch
from
150ff7e
to
62605ad
Compare
amesgen
force-pushed
the
blockfetch/milestone-1
branch
from
dd259c8
to
db26d58
Compare
Co-authored-by: Nicolas Bacquey <[email protected]>
facundominguez
force-pushed
the
blockfetch/milestone-1
branch
from
c07561d
to
dadb808
Compare
amesgen
requested review from
nfrisby,
jasagredo,
fraser-iohk,
dnadales and
RenateEilers
as code owners
dnadales
reviewed
Sep 9, 2024
| , gcLoEAndGDDConfig :: !(LoEAndGDDConfig LoEAndGDDParams) | ||
| } deriving stock (Eq, Generic, Show) | ||
|
|
||
| -- | Genesis configuration flags and low-level args, as parsed from config file or CLI |
There was a problem hiding this comment.
It might help to add comments and examples of these flags and their effect on the node's behaviour. It will surely help whoever integrates this with the CLI/Node.
| defaultCapacity = 100_000 -- number of tokens | ||
| defaultRate = 500 -- tokens per second leaking, 1/2ms | ||
| -- 3 * 2160 * 20 works in more recent ranges of slots, but causes syncing to | ||
| -- block in byron. |
There was a problem hiding this comment.
Suggested change
| -- block in byron. | |
| -- block in Byron. |
| -- values carry no special meaning. Someone needs to think about what values | ||
| -- would make for interesting tests. | ||
| gtLoPBucketParams = LoPBucketParams { lbpCapacity = 50, lbpRate = 10 }, | ||
| -- ^ REVIEW: Do we want to generate those randomly? |
There was a problem hiding this comment.
If we do not generate these randomly we should simply add a note that says that we could, but we should remove any pending REVIEWs from the code 🙏
| ChainSyncClientHandleCollection peer m blk -> | ||
| peer -> | ||
| m () | ||
| -- STM m (Maybe (peer, ChainSyncClientHandle m blk)) |
| -- Otherwise, the BlockFetch client would have to wait for | ||
| -- 'chainSelectionForFutureBlocks'. | ||
| -- | ||
| -- Note: we call 'chainSelectionForFutureBlocks' in all branches instead of |
Comment on lines
+600
to
+606
| -- REVIEW: What about all the threads that are waiting to write in the queue and | ||
| -- will write after the flush?! |
There was a problem hiding this comment.
In that case I'd say: let's add a note and remove the REVIEW item.
15 tasks
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Integrates a new implementation of the BulkSync mode, where blocks are downloaded from alternative peers as soon as the node has no more blocks to validate while there are longstanding requests in flight.
This PR depends on the new implementation of the BulkSync mode (IntersectMBO/ouroboros-network#4919).
cabal.projectis made to point to a back-port of the BulkSync implementation onouroboros-network-0.16.1.1.CSJ Changes
CSJ is involved because the new BulkSync mode requires to change the dynamo if it is also serving blocks, and it is not sending them promptly enough. The dynamo choice has an influence in the blocks that are chosen to be downloaded by BlockFetch.
For this sake, b93c379 gives the ability to order the ChainSync clients, so the dynamo role can be rotated among them whenever BlockFetch requests it.
b1c0bf8 provides the implementation of the rotation operation.
BlockFetch tests
c4bfa37 allows to specify in tests in which order to start the peers, which has an effect on what peer is chosen as initial dynamo.
c594c09 in turn adds a new BlockFetch test to show that syncing isn't slowed down by peers that don't send blocks.
Integration of BlockFetch changes
The collection of ChainSync client handles now needs to be passed between BlockFetch and ChainSync so dynamo rotations can be requested by BlockFetch.
The parameter
bfcMaxConcurrencyBulkSynchas been removed since blocks are not coordinated to be downloaded concurrently.These changes are in 6926278.
ChainSel changes
Now BlockFetch requires the ability to detect if ChainSel has run out of blocks to validate. This motivates 73187ba, which implements a mechanism to measure if ChainSel is waiting for more blocks (starves), and determines for how long.
The above change is not sufficient to measure starvation. The queue to send blocks for validation used to allow only for one block to sit in the queue. This would interfere with the ability to measure starvation since BlockFetch would block waiting for the queue to become empty, and the queue would quickly become empty after taking just 1 block. For download delays to be amortized, a larger queue capacity was needed. This is the reason why a fix to IntersectMBO/ouroboros-network#2721 was ported in 0d3fc28.
Miscellaneous fixes
CSJ jump size adjustment
When syncing from mainnet, we discovered that CSJ wouldn't sync the blocks from the Byron era. This was because the jump size was set to the length of the genesis window of the Shelley era, which is much larger than Byron's. When the jump size is larger than the genesis window, the dynamo will block on the forecast horizon before offering a jump that allows the chain selection to advance. In this case, CSJ and chain selection will deadlock.
For this reason we set the default jump size to the size of Byron's genesis window in 028883a. This didn't show an impact on syncing time in our measures. Future work (as part of deploying Genesis) might involve allowing the jump size to vary between different eras.
GDD rate limit
GDD evaluation showed an overhead of 10% if run after every header arrives via ChainSync. Therefore, in b7fa122 we limited how often it could run, so multiple header arrivals could be handled by a single GDD evaluation.
Candidate fragment comparison in the ChainSync client
We stumbled upon a test case where the candidate fragments of the dynamo and an objector were no longer than the current selection (both peers were adversarial). This was problematic because BlockFetch would refuse to download blocks from these candidates, and ChainSync in turn would wait for the selection to advance in order to download more headers.
The fix in e27a73c is to have the ChainSync client disconnect a peer which is about to block on the forecast horizon if its candidate isn't better than the selection.
Candidate fragment truncations
At the moment, it is possible for a candidate fragment to be truncated by CSJ when a jumper jumps to a point that is not younger than the tip of its current candidate fragment. We encountered tests where the jump point could be so old that it would fall behind the immutable tip, and GDD would ignore the peer when computing the Limit on Eagerness. This in turn would cause the selection to advance into potentially adversarial chains.
The fix in dc5f6f7 is to have GDD never drop candidates. When the candidate does not intersect the current selection, the LoE is not advanced. This is a situation guaranteed to be unblocked by the ChainSync client since it will either disconnect the peer or bring the candidate to intersect with the current selection.