I was wondering whether anyone has experienced a similar situation, and what a solution might be. Last night our scheduler performed a chain action. One of the tasks is a file action, (refresh and publish).
The file action has 3 main sources, combined in joins and append operations. The output only contained data from 2 of the 3 sources. When executed again this morning, the result was as expected containing data from all 3 sources.
This situation create a feeling of unstable processes, a feeling we desperately want to avoid ourselves but especially at the client side.
Alexander - Are all 3 sources copied onto the same server as running Omniscope? If one of the sources is accessed over a network from a different machine, a temporary networking failure can cause the source to be unavailable when Omniscope executes the Task List of Chain/File actions.
Locating all sources on the same machine, or ensuring redundant network access to all sources is the only way to minimise this risk.
Thanks. All sources that this datamanager is using are locally created and stored. Even if this server would go offline, all sources should be accessible as they are on the same harddisk.
Is it possible that, rather than the Merge block "skipping" an input, perhaps all 3 inputs were processed but one of them had zero records? You should be able to use a validation block above each merge input connector, to fail noisily if any inputs are produced incorrectly for whatever reason.
I understand what you mean, and we are using the validate block more often. This scripts has not been modified in the last 6 months, therefore it is strange that suddenly the behaviour of the script has changed in one run.
I checked all sources and all were (at the time of the task) as expected filled with data.
We have a new strange situation that we cannot explain. The scheduler was planned to execute a batch output that should make multiple PDF files. Yet only half of them was created, after this (not fully) executed task, the chain action continued with the following tasks. This is strange, as "propagate failure" is ticked.
We also send a bug report this afternoon, hope you can help us in investigating this issue. Thanks!