Commit Graph

2554 Commits

Author SHA1 Message Date
Justin Starry e3bdc38f0a
Add sanitized types for use in banking stage (#25067) 2022-05-11 00:30:48 +08:00
HaoranYi de96663cc4
fix typo (#25083) 2022-05-09 12:42:58 -05:00
Pankaj Garg 362b0526cd
Greedy receive in banking stage (#25060)
* Greedy receive in banking stage

* add upperbound to batch size and batching time

* update test_banking_stage_entryfication test
2022-05-08 10:47:55 -07:00
HaoranYi 8e37e364b1
fix typo in measure name (#25058) 2022-05-07 10:04:42 -05:00
Justin Starry c920d411f7
Clean up logging and make variables consistent (#25049) 2022-05-07 03:52:45 +08:00
Justin Starry 082502d4f3
Fail tx sanitization when ix program id uses lookup table (#25035)
* Fail tx sanitization when ix program id uses lookup table

* feedback
2022-05-07 03:19:50 +08:00
Christian Kamm cb6cd5d60f
FindPacketSenderStake: Improve metrics (#24971)
- separate names for vote and non-vote thread
- time unit postfixes (one is in ns!)
2022-05-06 21:16:13 +02:00
behzad nouri 492f89a170
checks account owner when initializing a vote-account (#25018)
A VoteAccount may only wrap an account if the account owner is
solana_vote_program:id or equivalently this check returns true:
solana_vote_program::check_id(account.owner())
2022-05-06 16:22:49 +00:00
behzad nouri a01291069a
initializes thread-pools with lazy_static instead of thread_local (#24853)
In addition to thread_local -> lazy_static change, a number of thread-pools are
initialized with get_max_thread_count to achieve parity with the older code in
terms of number of validator threads.
2022-05-05 20:00:50 +00:00
Justin Starry 7100f1c94b
Collect stats in streamer receiver and report fetch stage metrics (#25010) 2022-05-06 02:56:18 +08:00
carllin 870ac80b79
Prioritize BankingStage packets individually in min-max heap (#24187) 2022-05-04 21:50:56 -05:00
Christian Kamm 503d0baf6d
SigVerify: Add total time metrics for dedup/discard/verify (#24768)
* SigVerify: Add total time metrics for dedup/discard/verify

Previously it was impossible to determine the total time the stage spent
on these activities within a measurement window.

* SigVerify: Add _us postfix to time metrics
2022-05-03 14:59:25 +02:00
behzad nouri eff59193db
enforces that LAST_SHRED_IN_SLOT is also DATA_COMPLETE_SHRED (#24892)
A data shred cannot be LAST_SHRED_IN_SLOT if not also DATA_COMPLETE_SHRED.
So LAST_SHRED_IN_SLOT should also imply DATA_COMPLETE_SHRED:
https://github.com/solana-labs/solana/blob/74b586ae7/ledger/src/shredder.rs#L116-L117
https://github.com/solana-labs/solana/blob/74b586ae7/core/src/broadcast_stage/standard_broadcast_run.rs#L80-L81

However current shred constructs allow specifying a shred which is
LAST_SHRED_IN_SLOT but not DATA_COMPLETE_SHRED:
https://github.com/solana-labs/solana/blob/74b586ae7/ledger/src/shred.rs#L117-L118
https://github.com/solana-labs/solana/blob/74b586ae7/ledger/src/shred.rs#L272-L273

The commit updates ShredFlags so that if a shred is not
DATA_COMPLETE_SHRED it cannot be LAST_SHRED_IN_SLOT either.
2022-05-02 23:33:53 +00:00
apfitzge 112a0b475a
Revert "Refactor to use EpochSchedule from within RentCollector struct" (#24893)
* Revert "Ran cargo fmt"

This reverts commit 9052e41b32.

* Revert "Fix build error introduced by my editor setup, part 2"

This reverts commit 4dfeab3b38.

* Revert "Fix build error introduced by my editor setup"

This reverts commit 87fb78dc56.

* Revert "Remove redundant epoch_schedule from AccountsPackage"

This reverts commit c2f7f2fff8.

* Revert "Fix a test"

This reverts commit 36c0bdaa78.

* Revert "Fixes to initial code"

This reverts commit ed7813e698.

* Revert "Removing redundant EpochSchedule param from fns"

This reverts commit 5472d2e605.
2022-05-02 13:46:17 -05:00
behzad nouri e812430e28
defines shred flags using bitflags crate (#24874)
Shred flags uses raw bit-masking ops which lacks type-safety:
https://github.com/solana-labs/solana/blob/a829ddc92/ledger/src/shred.rs#L112-L114

This commit instead uses bitflags crate to define shred flags.
2022-05-01 19:25:15 +00:00
Pankaj Garg 88c16c0176
Check if quic is enabled before warming up quic connections (#24821)
* Check if quic is enabled before warming up quic connections

* fix after rebase

* don't start warmup service if quic not enabled

* fix test
2022-05-01 03:52:38 +00:00
Justin Starry a61652104b
Avoid holding lock guards in match expressions (#24805)
* Avoid holding bank forks read lock for RPC requests

* Avoid using lock guards in temporaries

* revert fetch stage change
2022-04-29 16:32:46 +08:00
Justin Starry 4e58b3870c
Update all BankForks methods to return owned values (#24801) 2022-04-28 18:51:00 +00:00
sakridge 5a430c15e2
Separate sigverify metrics for each verifier (#24744) 2022-04-28 01:16:17 -07:00
behzad nouri 0f60665100
replaces Shred::new_empty_coding with Shred::new_from_parity_shard (#24749)
Removing implementation details of shreds and payload offsets from
shredder, so that shredder does not need to mutate payload:
https://github.com/solana-labs/solana/blob/71ad12128/ledger/src/shred.rs#L968-L977

Also, Shred::new_from_data can simply obtain a slice as opposed to
Option<&[u8]>:
https://github.com/solana-labs/solana/blob/71ad12128/ledger/src/shred.rs#L268-L278
2022-04-27 18:04:10 +00:00
behzad nouri 081c844d6e
removes Shred::new_empty_data_shred (#24714)
Shred::new_empty_data_shred returns an invalid shred (i.e.
shred.sanitize() returns error). The method is only used in tests and
can be easily replaced with Shred::new_from_data. To keep the shred api
surface small, this commit removes this method.
2022-04-26 23:13:12 +00:00
behzad nouri 12ae8d3be5
returns Error when Shred::sanitize fails (#24653)
Including the error in the output allows to debug when Shred::sanitize
fails.
2022-04-25 23:19:37 +00:00
behzad nouri 895f76a93c
hides implementation details of shred from its public interface (#24563)
Working towards embedding versioning into shreds binary, so that a new
variant of shred struct can include merkle tree hashes of the erasure
set.
2022-04-25 12:43:22 +00:00
carllin 8a062273de
Move error counters to be reported by leader only at end of slot (#24581)
* Add error counters to leader metrics only

* Add dependencies
2022-04-23 18:10:47 -05:00
Michael Vines d0a8a16a57 ReplayStage no longer relies on Validator to reset the poh recorder at start 2022-04-22 21:17:49 -07:00
Michael Vines 84e3342612 Process blockstore after starting the TVU 2022-04-22 21:17:49 -07:00
Michael Vines 83e041299a Run real snapshot packager while processing blockstore at validator startup 2022-04-22 21:17:49 -07:00
HaoranYi 2d4defa477
fix typo (#24576) 2022-04-22 08:43:57 -05:00
Jon Cinque 0d51596224
sim: Override slot hashes account on simulation bank (#24543)
* sim: Override slot hashes during simulation

* Add simulation test program

* Address feedback

* Add AccountOverrides explicit type

* Cargo fmt
2022-04-22 12:32:31 +02:00
Justin Starry c544742091
Local cluster test cleanup and refactoring (#24559)
* remove FixedSchedule.start_epoch

* use duration for timing

* Rename to partition bool to turbine_disabled

* simplify partition config
2022-04-22 12:14:07 +08:00
Justin Starry 02bfb85c16
Refactor transaction processing in banking stage (#24336)
* Refactor transaction processing in banking stage

* feedback

* more feedback
2022-04-21 21:06:26 +08:00
Justin Starry d5127abf46
Only add hashes for completed blocks to recent blockhashes (#24389)
* Only add hashes for completed blocks to recent blockhashes

* feedback
2022-04-21 21:05:29 +08:00
Tao Zhu a21fc3f303
Apply transaction actual execution units to cost_tracker (#24311)
* Pass the sum of consumed compute units to cost_tracker

* cost model tracks builtins and bpf programs separately, enabling adjust block cost by actual bpf programs execution costs

* Copied nightly-only experimental `checked_add_(un)signed` implementation to sdk

* Add function to update cost tracker with execution cost adjustment

* Review suggestion - using enum instead of struct for CommitTransactionDetails
Co-authored-by: Justin Starry <justin.m.starry@gmail.com>

* review - rename variable to distinguish accumulated_consumed_units from individual compute_units_consumed

* not to use signed integer operations

* Review - using saturating_add_assign!(), and checked_*().unwrap_or()

* Review - using Ordering enum to cmp

* replace checked_ with saturating_

* review - remove unnecessary Option<>

* Review - add function to report number of non-zero units account to metrics
2022-04-21 07:38:07 +00:00
Justin Starry 79923c3b58
Refactor: Rename BlockhashQueue fields and methods for clarity (#24426) 2022-04-21 11:57:17 +08:00
HaoranYi d0761d0ca4
demote receive_window_num_slot_shreds to debug logging (#24505) 2022-04-20 08:51:46 -05:00
Michael Vines 05f32f287c solana-validator monitor now reports slot-level progress while loading blockstore 2022-04-19 22:09:48 -07:00
Michael Vines 9e4999ef6a Remove halt_at_slot from RuntimeConfig, it's not a runtime concern 2022-04-19 19:23:58 -07:00
Michael Vines 988210908c Move verify_udp_stats_access out of the way 2022-04-19 19:23:58 -07:00
Michael Vines c6f3da4879 blockstore_processor now accepts an Arc<Rwlock<BankForks>> 2022-04-19 19:23:58 -07:00
Michael Vines 0e2e0c8b7d Extract most storage-related services from the Tvu abstraction 2022-04-19 19:23:58 -07:00
Michael Vines 268a2109de Relocate hard forks info log 2022-04-19 19:23:58 -07:00
Michael Vines dd766042df Remove LedgerMetricReportService from TVU 2022-04-19 19:23:58 -07:00
behzad nouri 705ea53353
moves sign_shred and new_coding_shred_header out of Shredder (#24487) 2022-04-19 20:00:05 +00:00
Tao Zhu 94b0186a96
Cost model tracks builtins and bpf programs separately (#24468)
* Cost model tracks builtins and bpf programs separatele (enables adjusting block cost by actual bpf programs execution costs)

* Address reviews: expand test; add metrics stat
2022-04-19 13:25:47 -05:00
behzad nouri 3bbfaae7b6
moves shred stats to a separate file (#24484) 2022-04-19 18:25:09 +00:00
Jeff Washington (jwash) d9d0dad258
report swap mem as bytes like other metrics (#24455) 2022-04-19 10:03:25 -05:00
behzad nouri 039488b562
drops redundant turbine propagation path (#24351)
Most nodes in the cluster receive the same shred from two different
nodes: parent, and the first node of their neighborhood:
https://github.com/solana-labs/solana/blob/a8c695ba5/core/src/cluster_nodes.rs#L178-L197

Because of the erasure codings, half of the shreds are already
redundant. So this redundant propagation path will only add extra
overhead.

Additionally the very first node of the broadcast tree has 2x fanout
(i.e. 400 nodes) which adds too much load at one node.

This commit simplifies the broadcast tree by dropping the redundant
propagation path and removing the 2x fanout at root node.
2022-04-19 00:11:29 +00:00
behzad nouri 1d50832389
replaces counters with datapoints in gossip metrics (#24451) 2022-04-18 23:14:59 +00:00
Jason Davis c2f7f2fff8 Remove redundant epoch_schedule from AccountsPackage 2022-04-18 11:57:40 -05:00
Jason Davis 5472d2e605 Removing redundant EpochSchedule param from fns 2022-04-18 11:57:40 -05:00