-
Notifications
You must be signed in to change notification settings - Fork 2.1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
In-memory engine: implement core local array and collect read flow statistics #16945
Conversation
[REVIEW NOTIFICATION] This pull request has been approved by:
To complete the pull request process, please ask the reviewers in the list to review by filling The full list of commands accepted by this bot can be found here. Reviewer can indicate their review by submitting an approval review. |
Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au>
Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM. @tonyxuqqi any thoughts on RangeCacheOptions?
@afeinberg: Thanks for your review. The bot only counts approvals from reviewers and higher roles in list, but you're still welcome to leave your comments. In response to this:
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the ti-community-infra/tichi repository. |
@@ -417,6 +435,14 @@ impl Iterator for RangeCacheIterator { | |||
if self.valid { | |||
self.find_next_visible_key(true, guard); | |||
} | |||
|
|||
if self.valid { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
why not merge with the self.valid above.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Also the code is duplicated by many times, make it an shared inline function
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
after find_next_visible_key, the self.valid may not be true.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Please add a comment to mention that self.valid can be changed. It's not straightforward.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
But we should not have duplicated code when possible
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I have polished that.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Overall LGTM. Some details can be polished.
components/server/src/common.rs
Outdated
@@ -701,15 +704,15 @@ impl<T: fmt::Display + Send + 'static> Stop for LazyWorker<T> { | |||
|
|||
pub trait KvEngineBuilder: KvEngine { | |||
fn build( | |||
range_cache_engine_config: Arc<VersionTrack<RangeCacheEngineConfig>>, | |||
range_cache_engine_config: RangeCacheEngineOptions, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
the name and type does not match.
@@ -99,3 +104,21 @@ impl RangeCacheEngineConfig { | |||
} | |||
} | |||
} | |||
|
|||
pub struct RangeCacheEngineOptions { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The naming of RangeCacheEngineOptions is not appropriate. Statistics should not be part of Option.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think statistics should be just a member of RangeCacheMemoryEngine directly and initialized inside its constructor.
RangeCacheOptions is not necessary and it's also impropriate to include statistics as part of Options |
RocksDB put the |
/merge |
@SpadeA-Tang: It seems you want to merge this PR, I will help you trigger all the tests: /run-all-tests You only need to trigger
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the ti-community-infra/tichi repository. |
@SpadeA-Tang: In response to this:
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the ti-community-infra/tichi repository. |
/merge |
@SpadeA-Tang: It seems you want to merge this PR, I will help you trigger all the tests: /run-all-tests You only need to trigger
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the ti-community-infra/tichi repository. |
This pull request has been accepted and is ready to merge. Commit hash: 95e8ef6
|
…atistics (tikv#16945) ref tikv#16141 implement core local array and collect read flow statistics Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au>
* build: bump tikv pkg version (#17063) Signed-off-by: ti-chi-bot <ti-community-prow-bot@tidb.io> * resolved_ts: refactor log for the unexpected path (#17064) (#17068) ref #16818 Refactor logs for the unexpected path, print both the exsiting row and input key/value. Signed-off-by: cfzjywxk <lsswxrxr@163.com> Co-authored-by: cfzjywxk <lsswxrxr@163.com> * chore: add prow OWNERS files for critial configuration files (#17071) (#17108) close #17004 Signed-off-by: wuhuizuo <wuhuizuo@126.com> Co-authored-by: wuhuizuo <wuhuizuo@126.com> * OWNERS: Auto Sync OWNERS files from community membership (#16973) (#17120) Signed-off-by: Ti Chi Robot <ti-community-prow-bot@tidb.io> Signed-off-by: Jinpeng Zhang <zzzhangjinpeng@gmail.com> Co-authored-by: Jinpeng Zhang <zzzhangjinpeng@gmail.com> Co-authored-by: wuhuizuo <wuhuizuo@126.com> * In-memory engine: strip off read related code from engine.rs to a separte file (#16792) ref #16141 strip off read related code from engine.rs to a separte file Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * In-memory engine: collector iteration related metrics (#16735) ref #16141 collector iteration related metrics Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * In-memory engine: expose range cache engine configuration (#16791) ref #16141 expose range cache engine configuration Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * In-memory engine: do not load a range overlapped with an unresolved range (#16794) ref #16141 do not load a range overlapped with an unresolved range Co-authored-by: ti-chi-bot[bot] <108142056+ti-chi-bot[bot]@users.noreply.github.com> * server/node: rename Node to MultiRaftServer (#16783) close #16782 Signed-off-by: zhangjinpeng87 <zzzhangjinpeng@gmail.com> Signed-off-by: Jinpeng Zhang <zzzhangjinpeng@gmail.com> Co-authored-by: lucasliang <nkcs_lykx@hotmail.com> * In-memory engine: add metrics for memory usage (#16822) ref #16141 add metrics for memory usage Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> Co-authored-by: Jinpeng Zhang <zzzhangjinpeng@gmail.com> * In-memory engine: add metrics for gc duration, range load duration, and write duration (#16831) ref #16141 add metrics for gc duration, range load duration, and write duration Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * In-memory engine: delete lock physically rather than writting tombstone (#16852) ref #16141 delete lock physically rather than writting tombstone Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * In-memory engine: fix missing memory record when snapshot loading (#16850) ref #16141 fix missing memory record when snapshot loading Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * In-Memory Engine: add a tag to the range for easier troubleshooting (#16816) ref #16141 In-Memory Engine: add a tag into Range for diagnostics Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> Signed-off-by: Qi Xu <tonyxuqqi@outlook.com> Co-authored-by: SpadeA-Tang <u6748471@anu.edu.au> Co-authored-by: Qi Xu <tonyxuqqi@outlook.com> * In-memory engine: remove out of date pending ranges (#16872) ref #16141 remove out of date pending ranges Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * compile memory-engine by default and support dynamic config update for memory-engine (#16897) ref #16141 compile the in-memory-engine by default and support dynamic config update for memory-engine. The enabled flag would support enabled:true -> enabled:false and soft/hard limit also supports online change. But if enabled:false is set initially, the config change to make enabled:true needs the TiKV restart. Signed-off-by: tonyxuqqi <tonyxuqi@outlook.com> * In-memory engine: skiplist tombstone should be handled separetely in gc (#16944) ref #16141 skiplist tombstone should be handled separetely in gc Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * In-memory engine: manually implement partial_eq for CacheRange to avoid the consideration of tag (#16959) ref #16141 manually implement partial_eq for CacheRange to avoid the consideration of tag Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> Co-authored-by: tonyxuqqi <tonyxuqi@outlook.com> * In-memory engine: implement core local array and collect read flow statistics (#16945) ref #16141 implement core local array and collect read flow statistics Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * In-memory engine: fix seek does not set self.valid properly (#16991) ref #16141 fix seek does not set self.valid properly Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> Co-authored-by: ti-chi-bot[bot] <108142056+ti-chi-bot[bot]@users.noreply.github.com> * In-memory engine: add metrics for different types of iteration operations (#16997) ref #16141 add metrics for different types of iteration operations Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * In-memory engine: add metrics for range count (#16971) ref #16141 add metrics for range count Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> Signed-off-by: Spade A <71589810+SpadeA-Tang@users.noreply.github.com> Co-authored-by: lucasliang <nkcs_lykx@hotmail.com> * In-memory engine: fix gc worker not used (#17032) ref #16141 fix gc worker not used Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * raftstore: fix online config change panic for periodic-full-compact-start-time (#17069) close #17066 fix online config change panic for periodic-full-compact-start-time Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * config: test +HHMM offset in ReadableSchedule (#17081) close #17080 * Test both formats +HHMM offsets and as well as +HH:MM for ReadableSchedule. Signed-off-by: Alex Feinberg <alex@strlen.net> * In-memory engine: mvcc keys with different sequence number should only be handled once in GC (#17083) ref #16141, close #17060 mvcc keys with different sequen number should only be handled once in GC Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> Signed-off-by: Spade A <71589810+SpadeA-Tang@users.noreply.github.com> Co-authored-by: Neil Shen <overvenus@gmail.com> Co-authored-by: lucasliang <nkcs_lykx@hotmail.com> * In-memory engine: enable compaction filter for HybridEngine (#17087) ref #16141, close #17086 enable compaction filter for HybridEngine Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> Co-authored-by: ti-chi-bot[bot] <108142056+ti-chi-bot[bot]@users.noreply.github.com> * In-memory Engine: sequence number should be increased for each key (#17122) ref #16141, close #17114 sequence number should be increased for each key Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * In-memory engine: support changing the direction of iteration (#17129) ref #16141, close #17079 support reverse direction when iterating Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> Co-authored-by: ti-chi-bot[bot] <108142056+ti-chi-bot[bot]@users.noreply.github.com> * In-memory engine: clean lock cf tombstone in a background worker (#17128) ref #16141, close #17127 clean lock cf tombstone in a background worker Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * In-memory Engine: encode internal key trailer in little endian. (#17125) close #17082 An internal key trailer is `(sequence_number << 8) | value_type as u64` and it should be encoded in little endian to be consistent with RocksDB. Signed-off-by: Neil Shen <overvenus@gmail.com> * In-memory engine: set boundries when gc (#17145) ref #16141, close #17143 set boundries when gc Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> Co-authored-by: ti-chi-bot[bot] <108142056+ti-chi-bot[bot]@users.noreply.github.com> * In-memory engine: encode key for lock cf without mvcc version in delete range (#17142) ref #16141, close #17140 encode key for lock cf without mvcc version in delete range Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> Co-authored-by: ti-chi-bot[bot] <108142056+ti-chi-bot[bot]@users.noreply.github.com> * In-memory Engine: get GC safe_point from PD (#17144) close #17123 The `safe_point` is used to check if a cache is eligible to serve a read request. If the `safe_point` drifts into the future, for example by 10 minutes, the cache will not be able to serve any requests for those 10 minutes. This commit fixes such issue by getting the `safe_point` from PD. Signed-off-by: Neil Shen <overvenus@gmail.com> Co-authored-by: ti-chi-bot[bot] <108142056+ti-chi-bot[bot]@users.noreply.github.com> * In-Memory Engine: Eviction, Algorithmic loading (#17023) ref #16141, ref #16764 In-Memory Engine: Algorithmic Load and Eviction Signed-off-by: Alex Feinberg <alex@strlen.net> * In-memory engine: consider range overlaps when eviction (#17137) ref #16141, close #17131 consider range overlaps when eviction Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * In-memory engine: support delete range (#17151) ref #16141 support delete range Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * In-memory engine: evict range should consider loading range (#17154) close #17153 evict range should consider loading range Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> Co-authored-by: ti-chi-bot[bot] <108142056+ti-chi-bot[bot]@users.noreply.github.com> * In-memory engine: clear write batch when range load failed (#17156) ref #16141, close #17103 clear write batch when range load failed Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * In-memory engine: fix concurrency issue between delete range and write to memory (#17149) close #17104 fix concurrency issue between delete range and write to memory Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> Co-authored-by: ti-chi-bot[bot] <108142056+ti-chi-bot[bot]@users.noreply.github.com> * In-memory engine: add seek duration metrics (#17160) ref #16141 add seek duration metrics Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * In-memory engine: add metrics for prepare for write duration (#17161) ref #16141 add metrics for prepare for write duration Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * In-memory engine: set disk engine when start (#17165) ref #16141 set disk engine when start Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> Co-authored-by: ti-chi-bot[bot] <108142056+ti-chi-bot[bot]@users.noreply.github.com> * In-memory engine: avoid using snapshot cache (#16863) ref #16141 avoid using snapshot cache Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * In-memory engine: evict range after some raft commands (#17159) ref #16141 evict range when becomeing follower, merge, and deleting data ranges. Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * In-memory engine: `filter` plus one when any key is removed during GC (#17171) ref #16141 filter plus one when any key is removed during GC Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> Signed-off-by: Spade A <71589810+SpadeA-Tang@users.noreply.github.com> * periodic reload Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * avoid mem negative Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * make period of reload config Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * add overhead for delte entry size Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> * Fix missing field reload_period Signed-off-by: Neil Shen <overvenus@gmail.com> * Fix get_regions_in_range and BackgroundWorker compile Signed-off-by: Neil Shen <overvenus@gmail.com> --------- Signed-off-by: ti-chi-bot <ti-community-prow-bot@tidb.io> Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au> Signed-off-by: tonyxuqqi <tonyxuqi@outlook.com> Signed-off-by: Alex Feinberg <alex@strlen.net> Signed-off-by: Neil Shen <overvenus@gmail.com> Signed-off-by: Spade A <71589810+SpadeA-Tang@users.noreply.github.com> Co-authored-by: Ti Chi Robot <ti-community-prow-bot@tidb.io> Co-authored-by: cfzjywxk <lsswxrxr@163.com> Co-authored-by: wuhuizuo <wuhuizuo@126.com> Co-authored-by: Jinpeng Zhang <zzzhangjinpeng@gmail.com> Co-authored-by: Spade A <71589810+SpadeA-Tang@users.noreply.github.com> Co-authored-by: ti-chi-bot[bot] <108142056+ti-chi-bot[bot]@users.noreply.github.com> Co-authored-by: lucasliang <nkcs_lykx@hotmail.com> Co-authored-by: tonyxuqqi <tonyxuqi@outlook.com> Co-authored-by: SpadeA-Tang <u6748471@anu.edu.au> Co-authored-by: Qi Xu <tonyxuqqi@outlook.com> Co-authored-by: Alex Feinberg <alex@strlen.net>
What is changed and how it works?
Issue Number: Ref #16141
What's Changed:
This PR implements CoreLocalArray (which is implemented in RocksDB) for collecting metrics where, ideally, each thread only report stats to the cpu core executing it to readuce contention. In addition, this PR collects read flow and report to grafana like below:
![image](https://private-user-images.githubusercontent.com/71589810/328398169-257e9469-a16c-4e0c-b4e5-264da1cd1f49.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJnaXRodWIuY29tIiwiYXVkIjoicmF3LmdpdGh1YnVzZXJjb250ZW50LmNvbSIsImtleSI6ImtleTUiLCJleHAiOjE3MjU3NDU2MDUsIm5iZiI6MTcyNTc0NTMwNSwicGF0aCI6Ii83MTU4OTgxMC8zMjgzOTgxNjktMjU3ZTk0NjktYTE2Yy00ZTBjLWI0ZTUtMjY0ZGExY2QxZjQ5LnBuZz9YLUFtei1BbGdvcml0aG09QVdTNC1ITUFDLVNIQTI1NiZYLUFtei1DcmVkZW50aWFsPUFLSUFWQ09EWUxTQTUzUFFLNFpBJTJGMjAyNDA5MDclMkZ1cy1lYXN0LTElMkZzMyUyRmF3czRfcmVxdWVzdCZYLUFtei1EYXRlPTIwMjQwOTA3VDIxNDE0NVomWC1BbXotRXhwaXJlcz0zMDAmWC1BbXotU2lnbmF0dXJlPWNhMjk4ZTE5ZjNmNjBhM2YwMTBiYmQyMjFlYmEwYTBhYjg1YTcyZWY2M2U5ZGI1ZDJiNDQwZWM2OWM3MmI3YzgmWC1BbXotU2lnbmVkSGVhZGVycz1ob3N0JmFjdG9yX2lkPTAma2V5X2lkPTAmcmVwb19pZD0wIn0.BOy7-ZqUfRhoslek6D9RyUwapcdceO0Bn0CSQqdDh7M)
Related changes
pingcap/docs
/pingcap/docs-cn
:Check List
Tests
Side effects
Release note