DLT ERROR CODE: INGESTION_GATEWAY_CONNECTION_ERROR.SOURCE_CONNECTION_FAILURE

[iambdot]

we are using lakeflow to sync data from ms sql server to databricks, then 
we randomly started getting this connection error in our lakeflow pipeline. not sure how to resolve this

{
"id": "68dcb1f0-39af-11f1-aa8c-00163e2cf055",
"sequence": {
"data_plane_id": {
"instance": "execution",
"seq_no": 17763534305414123
},
"control_plane_seq_no": 17763436099082001
},
"origin": {
"cloud": "AWS",
"region": "us-east-2",
"org_id": 3454546387823487,
"pipeline_id": "79de0b5b-f1df-4caa-b7eb-0b90445e5925",
"pipeline_type": "INGESTION_GATEWAY",
"pipeline_name": "company-app-gateway-pipeline",
"cluster_id": "0416-140158-7lbda5n5",
"update_id": "9fb33b61-24a3-4ddf-bda8-2735b97120a9",
"request_id": "97434bb2-878f-448b-b6e0-a5a74d51d074",
"ingestion_source_type": "SQLSERVER"
},
"timestamp": "2026-04-16T16:14:56.655Z",
"message": "[DLT ERROR CODE: INGESTION_GATEWAY_CONNECTION_ERROR.SOURCE_CONNECTION_FAILURE] Update 9fbdfb has failed because of fatal error: Connection failure to the Source storage during operation.",
"level": "ERROR",
"error": {
"exceptions": [
{
"class_name": "tech.replicant.common.ExtractorException",
"message": "",
"stack": [
{
"declaring_class": "tech.replicant.db.jdbc.microsoft.SQLServerCDCExtractor",
"method_name": "fetchOperationsInternal",
"file_name": "SQLServerCDCExtractor.java",
"line_number": 1038
},
{
"declaring_class": "tech.replicant.db.jdbc.microsoft.SQLServerCDCExtractor",
"method_name": "fetchOperations",
"file_name": "SQLServerCDCExtractor.java",
"line_number": 527
},
{
"declaring_class": "tech.replicant.db.DBRealtimeExtractor",
"method_name": "fetchChanges",
"file_name": "DBRealtimeExtractor.java",
"line_number": 1465
},
{
"declaring_class": "tech.replicant.db.DBRealtimeExtractor$LogExtractorTask",
"method_name": "process",
"file_name": "DBRealtimeExtractor.java",
"line_number": 327
},
{
"declaring_class": "tech.replicant.db.DBRealtimeExtractor",
"method_name": "doRun",
"file_name": "DBRealtimeExtractor.java",
"line_number": 949
},
{
"declaring_class": "java.util.concurrent.FutureTask",
"method_name": "run",
"file_name": "FutureTask.java",
"line_number": 264
},
{
"declaring_class": "java.util.concurrent.Executors$RunnableAdapter",
"method_name": "call",
"file_name": "Executors.java",
"line_number": 539
},
{
"declaring_class": "java.util.concurrent.FutureTask",
"method_name": "run",
"file_name": "FutureTask.java",
"line_number": 264
},
{
"declaring_class": "java.util.concurrent.ThreadPoolExecutor",
"method_name": "runWorker",
"file_name": "ThreadPoolExecutor.java",
"line_number": 1136
},
{
"declaring_class": "java.util.concurrent.ThreadPoolExecutor$Worker",
"method_name": "run",
"file_name": "ThreadPoolExecutor.java",
"line_number": 635
},
{
"declaring_class": "java.lang.Thread",
"method_name": "run",
"file_name": "Thread.java",
"line_number": 840
}
],
"error_class": "INGESTION_GATEWAY_SOURCE_CONNECTION_FAILURE",
"message_parameters": {
"errorMessage": "tech.replicant.IStorage$StorageException: tech.replicant.jdbc.exception.JdbcSqlException: HikariPool-8 - Connection is not available, request timed out after 10000ms."
},
"short_message": ""
}
],
"fatal": true
},
"details": {
"update_progress": {
"state": "FAILED"
}
},
"event_type": "update_progress",
"maturity_level": "STABLE"
}

[balajij8]

The error indicates connection pool timeout. The Lake flow Ingestion Gateway cannot acquire a connection from the source SQL Server within the 10 second default window.

You can check on the below

• SQL Server Connection Exhaustion - Lake flow maintains persistent connections for log tailing. Validate if DBA recently lowered max degree of parallelism or reduced total user connections as both will cause issues - the gateway will be queued and timed out.
• Network & Gateway - As the pipeline is in us east 2, Validate if the SQL Server is on premises or in a completely different region as transient network jitter exceeding 10s will trigger this. Ensure that Databricks Ingestion Gateway is whitelisted in the SQL Server firewall and any intermediate NAT gateways are not hit by port exhaustion.
• CDC Lock Contention - Validate if intensive maintenance (index rebuilds or log truncation) is running on the source SQL server. The required CDC tables may be locked causing the extractor to hang until the pool times out.

You can increase the source connection configuration timeout to 30 s and verify source session availability after it.