1
0

[HUDI-3772] Fixing auto adjustment of lock configs for deltastreamer (#5207)

This commit is contained in:
Sivabalan Narayanan
2022-04-02 23:44:10 -07:00
committed by GitHub
parent cc3737be50
commit 84064a9b08
5 changed files with 102 additions and 33 deletions

View File

@@ -480,6 +480,12 @@ public class HoodieWriteConfig extends HoodieConfig {
.sinceVersion("0.11.0")
.withDocumentation("Control to enable release all persist rdds when the spark job finish.");
public static final ConfigProperty<Boolean> AUTO_ADJUST_LOCK_CONFIGS = ConfigProperty
.key("hoodie.auto.adjust.lock.configs")
.defaultValue(false)
.sinceVersion("0.11.0")
.withDocumentation("Auto adjust lock configurations when metadata table is enabled and for async table services.");
private ConsistencyGuardConfig consistencyGuardConfig;
private FileSystemRetryConfig fileSystemRetryConfig;
@@ -1968,6 +1974,9 @@ public class HoodieWriteConfig extends HoodieConfig {
* Hoodie Client Lock Configs.
* @return
*/
public boolean isAutoAdjustLockConfigs() {
return getBooleanOrDefault(AUTO_ADJUST_LOCK_CONFIGS);
}
public String getLockProviderClass() {
return getString(HoodieLockConfig.LOCK_PROVIDER_CLASS_NAME);
@@ -2443,6 +2452,11 @@ public class HoodieWriteConfig extends HoodieConfig {
return this;
}
public Builder withAutoAdjustLockConfigs(boolean autoAdjustLockConfigs) {
writeConfig.setValue(AUTO_ADJUST_LOCK_CONFIGS, String.valueOf(autoAdjustLockConfigs));
return this;
}
protected void setDefaults() {
writeConfig.setDefaultValue(MARKERS_TYPE, getDefaultMarkersType(engineType));
// Check for mandatory properties
@@ -2480,41 +2494,42 @@ public class HoodieWriteConfig extends HoodieConfig {
HoodieLayoutConfig.newBuilder().fromProperties(writeConfig.getProps()).build());
writeConfig.setDefaultValue(TIMELINE_LAYOUT_VERSION_NUM, String.valueOf(TimelineLayoutVersion.CURR_VERSION));
autoAdjustConfigsForConcurrencyMode();
}
private void autoAdjustConfigsForConcurrencyMode() {
boolean isMetadataTableEnabled = writeConfig.getBoolean(HoodieMetadataConfig.ENABLE);
// isLockProviderPropertySet must be fetched before setting defaults of HoodieLockConfig
final TypedProperties writeConfigProperties = writeConfig.getProps();
final boolean isLockProviderPropertySet = writeConfigProperties.containsKey(HoodieLockConfig.LOCK_PROVIDER_CLASS_NAME)
|| writeConfigProperties.containsKey(HoodieLockConfig.LOCK_PROVIDER_CLASS_PROP);
if (!isLockConfigSet) {
HoodieLockConfig.Builder lockConfigBuilder = HoodieLockConfig.newBuilder().fromProperties(writeConfig.getProps());
writeConfig.setDefault(lockConfigBuilder.build());
}
writeConfig.setDefaultOnCondition(!isLockConfigSet,
HoodieLockConfig.newBuilder().fromProperties(writeConfig.getProps()).build());
if (isMetadataTableEnabled) {
// When metadata table is enabled, optimistic concurrency control must be used for
// single writer with async table services.
// Async table services can update the metadata table and a lock provider is
// needed to guard against any concurrent table write operations. If user has
// not configured any lock provider, let's use the InProcess lock provider.
boolean areTableServicesEnabled = writeConfig.areTableServicesEnabled();
boolean areAsyncTableServicesEnabled = writeConfig.areAnyTableServicesAsync();
autoAdjustConfigsForConcurrencyMode(isLockProviderPropertySet);
}
if (!isLockProviderPropertySet && areTableServicesEnabled && areAsyncTableServicesEnabled) {
// This is targeted at Single writer with async table services
// If user does not set the lock provider, likely that the concurrency mode is not set either
// Override the configs for metadata table
writeConfig.setValue(WRITE_CONCURRENCY_MODE.key(),
WriteConcurrencyMode.OPTIMISTIC_CONCURRENCY_CONTROL.value());
writeConfig.setValue(HoodieLockConfig.LOCK_PROVIDER_CLASS_NAME.key(),
InProcessLockProvider.class.getName());
LOG.info(String.format("Automatically set %s=%s and %s=%s since user has not set the "
+ "lock provider for single writer with async table services",
WRITE_CONCURRENCY_MODE.key(), WriteConcurrencyMode.OPTIMISTIC_CONCURRENCY_CONTROL.value(),
HoodieLockConfig.LOCK_PROVIDER_CLASS_NAME.key(), InProcessLockProvider.class.getName()));
private void autoAdjustConfigsForConcurrencyMode(boolean isLockProviderPropertySet) {
if (writeConfig.isAutoAdjustLockConfigs()) {
// auto adjustment is required only for deltastreamer and spark streaming where async table services can be executed in the same JVM.
boolean isMetadataTableEnabled = writeConfig.getBoolean(HoodieMetadataConfig.ENABLE);
if (isMetadataTableEnabled) {
// When metadata table is enabled, optimistic concurrency control must be used for
// single writer with async table services.
// Async table services can update the metadata table and a lock provider is
// needed to guard against any concurrent table write operations. If user has
// not configured any lock provider, let's use the InProcess lock provider.
boolean areTableServicesEnabled = writeConfig.areTableServicesEnabled();
boolean areAsyncTableServicesEnabled = writeConfig.areAnyTableServicesAsync();
if (!isLockProviderPropertySet && areTableServicesEnabled && areAsyncTableServicesEnabled) {
// This is targeted at Single writer with async table services
// If user does not set the lock provider, likely that the concurrency mode is not set either
// Override the configs for metadata table
writeConfig.setValue(WRITE_CONCURRENCY_MODE.key(),
WriteConcurrencyMode.OPTIMISTIC_CONCURRENCY_CONTROL.value());
writeConfig.setValue(HoodieLockConfig.LOCK_PROVIDER_CLASS_NAME.key(),
InProcessLockProvider.class.getName());
LOG.info(String.format("Automatically set %s=%s and %s=%s since user has not set the "
+ "lock provider for single writer with async table services",
WRITE_CONCURRENCY_MODE.key(), WriteConcurrencyMode.OPTIMISTIC_CONCURRENCY_CONTROL.value(),
HoodieLockConfig.LOCK_PROVIDER_CLASS_NAME.key(), InProcessLockProvider.class.getName()));
}
}
}