When a connected system changes
Your configuration is unchanged, and still something stalls. A route connects two systems, and both are connected: SAP Business One on one side, the system behind the connection on the other. If one of them changes — a field in SAP Business One that the route reads from, or the shop's interface — Xchange notices it; the repair is up to you.
Such a change shows up in two very different ways: either the route keeps running and individual jobs fail — or the route is blocked and stands still. The way back into operation is different in each case.
The examples are two routes of Cambira Sanitärtechnik: "Kunden Webshop", whose data filter points to a field of the shop whose type changed with an update of the shop, and "Artikel Webshop", whose jobs the shop rejected for a while.
How to tell the two cases apart
Open Execution and look at the State column of the affected route.
| What it shows | What happened | Where to go next |
|---|---|---|
| Operating — and History shows failed jobs | The route is working; at the other end, individual records are rejected. | When the other system rejects the data |
| Paused (fault) | Xchange has stopped the route. It starts again only once the cause is fixed. | When automation is blocked |
In the second case, Xchange stops a route on its own. The reason is a prerequisite of the route that has changed — and as long as it differs, every run would be a run on wrong assumptions.
Two words for the same state. The list calls it Paused (fault), the route view says Automation blocked. Both mean the same thing. Paused (manual) is something else: the pause that someone set themselves.
When automation is blocked
The route "Kunden Webshop" fetches only the shop's trade customers into SAP Business One. Its
Data filter decides that: customer_group equals 20, the trade customer group. The condition
has a Type, here Text — Xchange measured it from the shop's field structure during setup and
stored it.
After an update of the shop, its field structure lists customer_group as a number. With that, the
stored value differs from reality: a text comparison against a field that holds a number misses the
records it is meant to hit — depending on the case, it hits too few or none at all. That is exactly
why Xchange steps in before the next run starts.
The sequence has five stages:
- In the source system, the type of a field that a filter condition points to changes.
- Xchange notices it during the regular comparison — or immediately, as soon as someone opens or saves the route.
- The route's automation is blocked. It no longer starts, neither timed nor triggered by an event.
- You adopt the new type on the marked conditions and activate the snapshot.
- The block lifts on its own, within six hours.
On the Execution screen, the block appears as a banner:

Adopt the new type
Follow the Open route link in the banner. It opens the route on the Source data tab, together
with the Messages. In the data filter, the Type cell is marked on every affected condition — on
"Kunden Webshop", the one condition on customer_group. The Messages list the finding as
"'Type' does not match the source schema — it now says 'Number'". As soon as you click the cell,
the hint names the type that the source system reports: "Source schema: 'Number'". Set Type to
this value.
If several conditions are affected, all of them are marked — Xchange names each one individually. Work through the whole list before you move on; an overlooked condition keeps the block in place.

Only activation counts
Your change first lands in the draft, like any other change — and a draft is at rest. The block lifts only once the changed snapshot is active.
The way there is the usual one: Change a running configuration describes it from the change to the activation.
The block lifts itself
After the activation, your part is done. The route runs again only after a while. Xchange checks the blocked routes again at fixed intervals, namely every six hours. On the next of these passes the types match, the block drops away, and the route resumes operation.
Six hours is the worst case: the first pass also always runs right after a restart of the Xchange service.
Leave the automation as it is. The switch on Execution works — the route counts as running afterwards — but the cause remains. The next comparison finds the same deviation again and blocks automation a second time. The switch is meant for the opposite case — for a route that you paused; it then shows Paused (manual).
When a field is gone entirely
A changed type is one way, a vanished field the other — and it ends differently. If the field is deleted or renamed in the source system, the check of the configuration snapshot reports "Field" does not exist in the field structure, with the name of the field in place of Field. The route keeps running regardless.
The consequence is still noticeable: a snapshot can be activated only once this finding is fixed. So you decide what the condition should mean from now on — select the new field or remove the condition — and then activate.
The same applies when a condition points through a relation to a linked target whose field structure has become unreadable: "Source schema missing". Remove the condition or make sure the field structure can be fetched again.
When the other system rejects the data
The second and much more common case looks quite different. The route "Artikel Webshop" had been running for months; one Monday morning the shop rejected every transferred item, because it required a field that was missing from Xchange's data. The route kept running — it did its work, and the other end rejected it. The failed jobs piled up in Clarification, all with the same cause.
The case was fixed in the shop — at 1:40 PM the field was optional there again. Routes and configuration snapshot stayed as they were. After that, all that was left was to catch up on the jobs that had been left behind.
That is the normal case when the other system rejects data. So before you change anything in your configuration, clarify: Is the change over there intended, and will it stay? Then adjust your route. Was it a mistake, or will it be reverted? Then wait and catch up afterwards — otherwise a changed and activated route would be double work.
How to catch up on the jobs left behind is described in Retry failed jobs; how to decide a single case, in Resolve a clarification.
What comes next
- You do not yet know whether another system is the cause at all: It ran for months and stopped yesterday.
- The jobs from the time of the disruption are still pending: Retry failed jobs.
- You want to push a single record immediately, before the next run: Push manually.
See also
- Routes — Source data (reference) — the data filter and the change filter with all their settings
- Routes — Automation (reference) — what the automation switch does and where its effect ends
- Execution (reference) — all states a route can take in the list
- Configuration Snapshots (core concept) — why a change takes effect only with the activation