SSIS Reruns
Overview
A package that fails and is immediately rerun shows as one failure and one success, and almost every piece of reporting in an estate nets that to fine.
The number people actually want is how often it worked without somebody helping, and that number is on no page anywhere today.
Where to find it
Expand a server in the tree, expand a database holding an Integration Services catalog, then SSIS → Reruns.
Three shapes, kept apart
| Shape | What it is |
|---|---|
| Retry after failure | Another run started shortly after this one failed. An Agent retry, or a person who noticed. Either way the package did not work on its own. |
| Rerun after success | Another run started shortly after this one succeeded. Somebody who did not trust the first one, or a double trigger. Both are worth knowing about. |
| Worked on attempt three | A chain of failures ending in a success. The dashboard is green and the package is not. |
The window decides the answer
Which is why it is on the toolbar rather than buried in the code.
| Setting | What it catches |
|---|---|
| Within 10 min | The tight Agent retry. |
| Within 30 min | The Agent retry and the person who noticed. |
| Within 2 hours | Catches the next scheduled run and calls it a retry, which would report every hourly package as permanently broken. |
Thirty minutes is the default for that reason.
Reading the grid
| Column | What it is |
|---|---|
| Package / Project | The package. |
| Executions | Total runs in the window. |
| Worked | Runs that succeeded first time, with nothing before them. |
| Eventually | Runs that succeeded only after one or more failures. |
| Never | Failures with no successful attempt after them. |
| Ran twice | Reruns that followed a success. |
| Clean rate | Worked divided by attempts. This is the number the page exists to produce. |
| Worst | The longest chain of attempts before a success. |
| State | Reliable, needs a retry to work, or green on a dashboard and broken in practice. |
The toolbar
| Control | Options | Default |
|---|---|---|
| Window | 7 days, 30 days, 90 days | 30 days |
| Rerun gap | Within 10 min, Within 30 min, Within 2 hours | Within 30 min |
| Failures | Opens SSIS Package Failures. |
Where the data comes from
catalog.executions: the package, project, start, end and status of every run, with the runs of one package put in time order so the chains can be found.
Related reports
| Report | Why you would go there |
|---|---|
| SSIS Package Failures | What the failed attempts actually said. |
| SSIS Job Disagreement | Whether the Agent job reported the failure at all. |
| SSIS Package Speed | Whether the failing attempts are fast failures, which skew durations. |
| SSIS Schedule | Whether the “retry” is actually the next scheduled run. |
| SSIS Self Contention | Whether the first attempt fails because of what runs beside it. |
Frequently asked questions
A clean rate of 60 percent on a package our dashboard shows as green. That is the finding. Four runs in ten needed a retry, and the retry succeeded, so nothing downstream ever noticed.
Every hourly package shows as retried. Your rerun gap is too wide. At two hours, the next scheduled run of an hourly package looks exactly like a retry. Drop to thirty or ten minutes.
What counts as a rerun after success? A run that started within the gap of a run that succeeded. Sometimes a double trigger, sometimes somebody who did not trust the first one, and worth a look either way.
Should I remove the Agent retries? Not necessarily. What the page gives you is the honest reliability number underneath them, so the decision is informed rather than blind.