What this solves#
A rollout ended Partial completed. 1,180 of 1,200 devices received the files. You need to send again.
But "send again" is four different things. Pick the wrong one and the 1,180 devices that already have the file get it a second time. The overnight link saturates again and devices overwrite files that were fine.
The four differ in what they send. This document covers what each one sends and when to pick it.
At a glance#
| Action | What it sends | Where | When it appears |
|---|---|---|---|
Send again |
The whole run. Including what already arrived | The ... in the run detail panel's header |
Any finished run — Completed, Partial completed, Failed, Stopped |
Retry failed files |
Only the files that failed | The same ... |
Failed and Partial completed rows only |
Resend missing files |
Only what is not at the destination now | The Assurance side | Rows where Stored at destination is Violated |
Per-file Retry |
One file, or one folder | The file list in the transfer detail | On failed rows |
Using
Send againon a rollout where 1,180 devices already have the file sends it to all 1,200 again. The second row of this table is the right choice.
Who this is for#
| Role | When |
|---|---|
| Operations | An overnight rollout finished only partly |
| System administrator | One device keeps failing |
| Business users | A single file did not arrive |
One at a time#
1. Send again — the whole run, from the start#
Press the run row and a detail panel opens on the right. The ... in that panel's header holds it. The row itself has no ... - this is where people look first and find nothing. It is always there for a finished run, whatever its status.
It sends every file again, so the product asks for confirmation first. Confirm and a new run is created.
Pick this when
- The source changed and everything must be re-sent
- Files were deleted or corrupted at the destination
- The run itself has to be repeated to produce a fresh result
Do not pick this when
- Only part failed. Everything that succeeded goes again
- There are many targets and the link is narrow
2. Retry failed files — only what failed#
Same menu, but it appears only on Failed and Partial completed rows. A fully successful run has nothing failed to retry, so the item is not there.
Starting it raises the notice "Retrying the failed files."
Pick this when
- A rollout or collection ended
Partial completed. The most common case - Some devices were offline and have come back
- A few files were held up by low storage or a file lock
3. Resend missing files — only what the destination lacks now#
On the Assurance side, and only on rows where Stored at destination is Violated.
This one is different from the first two. The server checks the destination again and sends only what is not there. It reads the current state, not the transfer record.
It takes a moment, and the answer comes as a notice.
| Notice | Meaning |
|---|---|
Sending 12 missing files. |
12 are missing and are being sent |
Nothing is missing at the destination. |
Everything is there. No run starts |
Either, followed by Only the first N files were checked. |
The server checked only part of the run. "Nothing is missing" is not a check of every file |
Pick this when
- The transfer says success but files are not visible at the destination
- Someone deleted files on the target side
- A collection finished but the aggregate is short
Caution: if Only the first N files were checked. appears, it was not an exhaustive check. On runs with very many files, always look for that sentence.
4. Per-file Retry — just that one#
In the transfer detail's file list, press Retry on a failed row. It works on a single file or on a folder.
The product enforces two things:
- One at a time. While a retry is in flight, another is not accepted
- Pressing twice does not send twice. The row's
Retrystays disabled until the list has refreshed
If the device has been removed or is disconnected, the retry does not start, because it could only be refused.
Pick this when
- A business user asks for "just this one file"
- You are working through failures one by one, reading why each failed
What the ... menu offers#
This menu is in the run detail panel's header, not on the run row. Press the row, the panel opens, and the ... at the right of its header is the menu below.
The menu changes with the run's state. Actions that run cannot perform are not shown at all. A line splits the menu in two - above it, what moves the run; below it, what you look at or change on the Flow.
| Run state | Top group (move it) | Bottom group (look and change) |
|---|---|---|
Running |
Pause, Stop |
View transfer detail, Edit flow |
Recovering |
Pause, Stop. The same as Running |
View transfer detail, Edit flow |
Paused |
Resume, Stop |
View transfer detail, Edit flow |
Planned (scheduled, not started) |
Run now, Skip this run |
Edit flow only |
Skipped |
Run now |
Edit flow only |
Completed |
Send again |
View transfer detail, Edit flow |
Partial completed |
Send again, Retry failed files |
View errors, Edit flow |
Failed |
Send again, Retry failed files |
View errors, Edit flow |
Stopped |
Send again |
View transfer detail, Edit flow |
Four things save confusion:
Recoveringis a run in flight. It is retrying, not waiting to start, so the menu is theRunningone:PauseandStop. Stopping it ends the run and its retries with it.- A run that has not started has nothing to look at.
PlannedandSkippedhave no transfer record yet, so the detail entry is absent. Edit flowis the last item in every state. Even when nothing else is offered, that is there.PauseandStopare not decided by the state alone. The product also asks whether that run can be held right now, so a running transfer may show only one of the two. If it is not there, that run cannot do it right now - nothing is broken.
Skipping this occurrence of a scheduled run leaves the Flow's schedule intact. The next occurrence runs normally.
A run stopped once it had already reached 100% reads as
Completed, and its menu matchesCompleted.
Resending one step or one target#
A row in the run list is not only the Flow row. Numbered child rows sit under it:
- For a chained Flow, one step
- For a distribution, one target device
- For a collection, one source device
A child row's menu acts on that row alone. To resend to one device out of 1,200, use that device's row, not the Flow row.
Choosing by situation#
| Situation | Pick | Why |
|---|---|---|
| 1,200-store rollout, 20 failed | Retry failed files |
Sending to 1,180 again wastes the link |
| A fault was found in the source after the rollout | Send again |
Devices that already have it must be overwritten with the fixed file |
| Collection succeeded but the aggregate folder is short | Resend missing files |
You need the current destination, not the transfer record |
| One factory keeps failing | That device's row menu | Leaves the other factories alone |
| A user asks for one drawing | Per-file Retry |
The smallest unit |
| Yesterday's run is unfinished and today's schedule fires | Skip this run | Do not overlap |
Checking the result#
A resend creates a new run row. The original row stays. You need both to see the whole picture.
A clean outcome looks like:
- The new row reads
Completed - The failed count is 0
Contract Statusin Assurance isSatisfied
If it ends Partial completed again, the same device failed again. Do not keep resending. Look at that device, and see "The same issue every day" in Decide in Advance What Happens When a Transfer Goes Wrong.
Errors and exceptions#
| What you see | Meaning | What to do |
|---|---|---|
No Retry failed files in the menu |
That run has no failed files | If it is Completed but files are missing at the destination, use Resend missing files |
Nothing is missing at the destination. |
Everything is there | No transfer starts. This is normal |
Followed by Only the first N files were checked. |
Not an exhaustive check | On runs with many files, cross-check another way |
Per-file Retry does not respond |
The device was removed or is disconnected | Restore the device connection first |
Could not send the missing files. |
The request was refused | Check permissions and server state |
Send again asks for confirmation |
It will send every file again | Make sure that is what you want |
Operating notes#
- Do not use
Send againonPartial completed. It is the most common mistake and the most expensive in bandwidth. - Find out why it failed before resending. Resending to a full disk fails again.
- A skipped device is not a failure. Devices skipped by policy do not appear under
Retry failed files. Read theSkippedcount underDevicesand resend to those separately. - Do not collide with automatic retries. An issue set to
Keep retryingmay still be trying. Act after the run ends. Resend missing filesre-reads the destination. With many targets the answer takes a moment. Do not press twice.
By industry#
| Industry | How it is used |
|---|---|
| Retail | One Retry failed files in the morning after the overnight rollout, filling the missed stores before opening |
| Manufacturing | Retry from the line's child row, leaving other lines' production alone |
| Finance | Resend missing files after branch collection to catch aggregate gaps. Do not trust the transfer record alone |
| Healthcare | Retry per inspection device, and per-file Retry is used often |
| Broadcast | Close to air time, resend only what failed. There is no time for the rest |
See also#
- Reading Per-Device Results and What Is Missing — find what failed first
- Decide in Advance What Happens When a Transfer Goes Wrong — reduce failures in the first place
- Define Transfer Success with an Assurance Contract — the
Stored at destinationcondition - Automatically Resume File Transfers After a Network Interruption