fix pipeline settings

This commit is contained in:
etreschenkov 2025-11-26 16:08:17 +03:00
parent 930a563172
commit d4f689e85a
6 changed files with 49 additions and 99 deletions

View file

@ -75,19 +75,29 @@ ct.encoding=windows-1251
#----------------------------Pipeline settings----------------------------#
#???-?? ????????? ?? xml ?? ???? ????????????
#batch size for saving in phone_fid_map
ct.saver-subject-batch-size=15000
ct.pipeline.subject-batch-size=5000
#????????? ? PHONE ? NAME + ?????? ? ???? SIGNALS_PHONE_FID_MAP
#limit on select phone_fid_map
ct.pipeline.subject-batch-size=10000
#STAGES SETTINGS
#select NAME + FID BY PHONE
ct.pipeline.name-enricher-stage-concurrency=20
#????????? ? ID ? ?????? ????? xml ??? ??. ???????
ct.pipeline.name-enricher-queue-capacity=20
#select ID BY FID
ct.pipeline.id-enricher-stage-concurrency=10
#?????????? ??????? ??? ??????
ct.pipeline.id-enricher-queue-capacity=20
#Save part of signal xml file and merge; only one thread
ct.pipeline.merger-queue-capacity=50
#select and calc flags; instant run after name-enricher
ct.pipeline.flag-enricher-stage-concurrency=10
ct.pipeline.flag-enricher-queue-capacity=50
#Backoff settings
ct.pipeline.error-attempt-count=5
#????? ???????? ????? ????????? ???????
ct.pipeline.error-attempt-delay=2s
#??????? ?? ???? ???????
ct.pipeline.retryable-exceptions=org.springframework.dao.DataAccessException,org.springframework.dao.DataIntegrityViolationException
#Pipeline global settings
ct.pipeline.global-timeout=2h
#???-?? ??????????? ???????????? ?????????????? ????????
ct.pipeline.parallel-requests=3
ct.pipeline.retryable-exceptions=org.springframework.dao.DataAccessException,org.springframework.dao.DataIntegrityViolationException
ct.pipeline.parallel-requests=3

View file

@ -6,9 +6,13 @@ import org.springframework.boot.context.properties.bind.DefaultValue;
public record PipelineSettings(
@DefaultValue("5000") Integer subjectBatchSize,
@DefaultValue("3") Integer saveSubjectStageConcurrency,
@DefaultValue("15") Integer nameEnricherStageConcurrency,
@DefaultValue("20") Integer nameEnricherQueueCapacity,
@DefaultValue("10") Integer idEnricherStageConcurrency,
@DefaultValue("15") Integer idEnricherQueueCapacity,
@DefaultValue("10") Integer flagEnricherStageConcurrency,
@DefaultValue("100") Integer flagEnricherQueueCapacity,
@DefaultValue("50") Integer mergerQueueCapacity,
@DefaultValue("5") Integer errorAttemptCount,
@DefaultValue("2s") Duration errorAttemptDelay,
@DefaultValue("5m") Duration globalTimeout,

View file

@ -2,30 +2,26 @@ package ru.nbch.credit_tracker.config.task;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
import ru.nbch.credit_tracker.config.settings.CreditTrackerSettings;
import ru.nbch.credit_tracker.model.impl.PartFilePayload;
import ru.nbch.credit_tracker.model.impl.SubjectPayload;
@Configuration
public class StageConfig {
@Bean("saveSubjectQueueFactory")
public StageQueueFactory<SubjectPayload> saveSubjectQueueFactory() {
return new StageQueueFactory<>(50);
}
@Bean("nameEnricherQueueFactory")
public StageQueueFactory<SubjectPayload> nameEnricherQueueFactory() {
return new StageQueueFactory<>(10);
public StageQueueFactory<SubjectPayload> nameEnricherQueueFactory(CreditTrackerSettings creditTrackerSettings) {
return new StageQueueFactory<>(creditTrackerSettings.getPipeline().nameEnricherQueueCapacity());
}
@Bean("idEnricherQueueFactory")
public StageQueueFactory<SubjectPayload> idEnricherQueueFactory() {
return new StageQueueFactory<>(10);
public StageQueueFactory<SubjectPayload> idEnricherQueueFactory(CreditTrackerSettings creditTrackerSettings) {
return new StageQueueFactory<>(creditTrackerSettings.getPipeline().nameEnricherQueueCapacity());
}
@Bean("partFileQueueFactory")
public StageQueueFactory<PartFilePayload> partFileQueueFactory() {
return new StageQueueFactory<>(100);
public StageQueueFactory<PartFilePayload> partFileQueueFactory(CreditTrackerSettings creditTrackerSettings) {
return new StageQueueFactory<>(creditTrackerSettings.getPipeline().mergerQueueCapacity());
}
}

View file

@ -9,6 +9,7 @@ import java.util.concurrent.TimeUnit;
import java.util.function.BiFunction;
import java.util.function.Supplier;
import lombok.extern.slf4j.Slf4j;
import org.springframework.beans.factory.InitializingBean;
import org.springframework.beans.factory.annotation.Qualifier;
import org.springframework.stereotype.Component;
import ru.nbch.credit_tracker.config.settings.CreditTrackerSettings;
@ -30,7 +31,7 @@ import ru.nbch.credit_tracker.utils.error.ExceptionUtils;
@Component
@Slf4j
public class PipelinePackageManager {
public class PipelinePackageManager implements InitializingBean {
private final SignalRestService signalRestService;
private final FlagsEnricher flagsEnricherV2;
private final StageQueueFactory<SubjectPayload> idEnricherQueueFactory;
@ -48,11 +49,7 @@ public class PipelinePackageManager {
return t;
});
private final ExecutorService subjectSaverExecutor = Executors.newFixedThreadPool(3, r -> {
Thread t = new Thread(r);
t.setName("subject-saver-%d".formatted(t.threadId()));
return t;
});
private final ExecutorService subjectSaverExecutor;
private final Semaphore semaphore;
private final PipelineBackoff pipelineBackoff;
private final Saver saver;
@ -80,9 +77,13 @@ public class PipelinePackageManager {
this.fidNameEnricher = fidNameEnricher;
this.creditTrackerSettings = creditTrackerSettings;
this.semaphore = new Semaphore(creditTrackerSettings.getPipeline().parallelRequests());
log.info("pipeline settings: {}", creditTrackerSettings.getPipeline());
this.pipelineBackoff = pipelineBackoff;
this.saver = saver;
this.subjectSaverExecutor = Executors.newFixedThreadPool(creditTrackerSettings.getPipeline().parallelRequests(), r -> {
Thread t = new Thread(r);
t.setName("subject-saver-%d".formatted(t.threadId()));
return t;
});
}
public void buildAndRun(Long packageId, Path pkgPath) throws Exception {
@ -147,4 +148,9 @@ public class PipelinePackageManager {
throw e;
}
}
@Override
public void afterPropertiesSet() throws Exception {
log.info("pipeline settings: {}", creditTrackerSettings.getPipeline());
}
}

View file

@ -12,6 +12,7 @@ import java.util.concurrent.TimeUnit;
import java.util.function.Consumer;
import lombok.extern.slf4j.Slf4j;
import org.springframework.stereotype.Service;
import ru.nbch.credit_tracker.config.settings.CreditTrackerSettings;
import ru.nbch.credit_tracker.entities.app.FlagDetail;
import ru.nbch.credit_tracker.mapper.signal.PhoneFidMapMapper;
import ru.nbch.credit_tracker.model.SubjectProfile;
@ -26,12 +27,12 @@ public class FlagsEnricher implements Consumer<SubjectPayload> {
private final ExecutorService ioPool;
public FlagsEnricher(FlagsService2 flagsService,
PhoneFidMapMapper phoneFidMapMapper) {
CreditTrackerSettings creditTrackerSettings) {
this.flagsService = flagsService;
this.ioPool = new ThreadPoolExecutor(10,
this.ioPool = new ThreadPoolExecutor(creditTrackerSettings.getPipeline().flagEnricherStageConcurrency(),
10,
0L, TimeUnit.MINUTES,
new LinkedBlockingQueue<>(100),
new LinkedBlockingQueue<>(creditTrackerSettings.getPipeline().flagEnricherQueueCapacity()),
r -> {
Thread t = new Thread(r);
t.setName("flags-enricher-%d".formatted(t.threadId()));

View file

@ -1,67 +0,0 @@
package ru.nbch.credit_tracker.service.task.impl;
import java.util.List;
import lombok.extern.slf4j.Slf4j;
import org.springframework.stereotype.Service;
import ru.nbch.credit_tracker.config.settings.CreditTrackerSettings;
import ru.nbch.credit_tracker.entities.app.PhoneFidMap;
import ru.nbch.credit_tracker.enums.Status;
import ru.nbch.credit_tracker.mapper.signal.UserPackageMapper;
import ru.nbch.credit_tracker.model.SubjectProfile;
import ru.nbch.credit_tracker.model.impl.SubjectPayload;
import ru.nbch.credit_tracker.service.indic.PhoneService2;
import ru.nbch.credit_tracker.service.task.Handler;
/**
* Стадия для сопоставления FID к номеру телефона и регистрации пакета в базе данных.
* Так же выполняется поиск и обогащение перс.данными из NAME для дальнейшей обработки.
*/
@Service
@Slf4j
public class SaveSubjectHandler implements Handler<SubjectPayload, SubjectPayload> {
private final PhoneService2 phoneService2;
private final UserPackageMapper userPackageMapper;
private final CreditTrackerSettings creditTrackerSettings;
public SaveSubjectHandler(PhoneService2 phoneService2,
UserPackageMapper userPackageMapper,
CreditTrackerSettings creditTrackerSettings) {
this.phoneService2 = phoneService2;
this.userPackageMapper = userPackageMapper;
this.creditTrackerSettings = creditTrackerSettings;
}
@Override
public SubjectPayload handle(SubjectPayload taskPayload) {
List<SubjectProfile> subjects = taskPayload.getSubjects();
Long packageId = taskPayload.getPackageId();
log.debug("SaveSubjectStage handle by packageId: {}, size of bucket for save: {}", taskPayload.getPackageId(), subjects.size());
List<PhoneFidMap> subjectForSave = subjects.stream().map(subject -> PhoneFidMap.builder()
.subjectId(subject.getIdentifier())
.packageId(packageId)
.phone(subject.getPhone())
.build()).toList();
List<Long> ids = phoneService2.registerSubjects(subjectForSave);
for (int i = 0; i < subjects.size(); i++) {
subjects.get(i).setMapId(ids.get(i));
}
return taskPayload;
}
@Override
public void handleOnPoison(SubjectPayload poison) {
log.trace("Got poison; set package on PROCESSING by packageId: {}", poison.getPackageId());
userPackageMapper.updatePackageStatus(poison.getPackageId(), Status.PROCESSING.name());
}
@Override
public int preferredConcurrency() {
return creditTrackerSettings.getPipeline().saveSubjectStageConcurrency();
}
@Override
public SubjectPayload cookPoison() {
return SubjectPayload.POISON;
}
}