In Informatica PowerCenter, failed rows land in different places depending on how the session is configured. By default, transformation errors go to the session log and rejected rows go to the reject file. If you turn on row error logging, errors go to error tables or an error file instead, and the reject file is no longer created. With full pushdown optimization, the database handles errors, and the Integration Service does not write reject rows to the reject file. To handle errors well, choose one logging method, read the session log from the first error, set a recovery strategy, and use the Debugger for logic problems. PowerCenter 10.5.x standard support also ended on March 31, 2026, so decide how long you will keep investing in it.
Most PowerCenter troubleshooting advice is a list of generic tips. The questions that cost teams time are more specific: where did my rejected rows go, why will this session not recover, and which log should I read first? This page answers those, and then covers what the 2026 support date means for how much effort you put into old workflows.
If you run a lot of PowerCenter, our ETL operations and workflow efficiency articles cover the surrounding practices.
| Situation | Where Failed Rows or Messages Land | Watch Out For |
|---|---|---|
| Default session | Transformation errors in the session log, rejected rows in the reject file | Easy to miss if nobody reads the files |
| Row error logging to tables | Error tables, including the source row data | No reject file is created, and dropped rows are not written to the session log |
| Row error logging to a file | An error log file | Same replacement of the reject file |
| Verbose data tracing | Row data in the session log, which can also be added alongside row error logging | Slows the session and enlarges logs |
| Full pushdown optimization | Handled by the database | The Integration Service does not write reject rows to the reject file |
| XML sources | The session log | Row errors from XML sources cannot be logged to error tables |
What this means for you: Before a failure forces the question, decide which method each session uses and write it down.
Message codes carry a short prefix that points to the area, so note the prefix when you search Informatica’s documentation or ask for support.
When you enable relational error logging, PowerCenter creates error tables the first time the session runs. Informatica documents these tables:
If you use the same table prefix across sessions, you can collect row errors from many sessions into one set of tables, and a session parameter in a parameter file keeps that manageable. Two cautions apply. First, the error tables store the source row data, so treat them as sensitive. Second, row errors from XML sources cannot be logged this way and appear in the session log.
What this means for you: Agree on one table prefix and one retention rule, then query the tables as part of daily monitoring.
Informatica offers three recovery strategies for a session: resume from the last checkpoint, restart the task, or fail the task and continue the workflow.
To recover, right-click the session in the Workflow Monitor and choose Recover Task, or use the pmcmd starttask command with the recover option. Test recovery on a non-production run before you depend on it, and pair it with incremental loads and duplicate checks. Our automated ETL testing and data migration testing tools articles describe how to prove that a rerun leaves the target correct.
Decision tasks can branch on session status, email tasks can notify the right team, and command tasks can trigger a recovery script. Standardize these across workflows, store run history in a logging database, and review error trends weekly. Our ETL data pipeline automation article shows how teams extend this into broader automation.
Informatica’s own site says PowerCenter 10.5.x standard support ended on March 31, 2026, and third-party sources report extended support to about March 2027, so confirm the dates for your version. Salesforce completed its acquisition of Informatica in November 2025. None of this means workflows stop running, but it changes how much to invest in them.
A sensible approach is to fix the errors that cause real risk now, document the rest, and decide a direction:
AI assistants will increasingly use the data these pipelines deliver, so tested, traceable loads matter more than before. That is an expectation, not a certainty, but clean error handling today makes any later migration easier. If you are also weighing Oracle tools, see our ODI vs Informatica comparison, and for general practices, see our article on maximizing data integration success with Informatica.
This section is planning guidance, not a statement of DataTerrain’s authorization status or government work. Public sector and other regulated teams should check these points:
The costly mistake is spending months polishing workflows without deciding how long they will run. DataTerrain brings 17+ years and 400+ clients in data and reporting modernization, with PowerCenter stabilization, upgrade or migration planning, automated validation, and operating handover planned together.
Informatica PowerCenter and MDM | Advantages of Informatica PowerCenter | Mastering Data Quality with Informatica | Informatica Cloud vs SnapLogic | SnapLogic vs Informatica | AWS Glue vs Informatica Cloud