I've been beating my head against the wall for the past week trying to figure out why my workflow is failing 10-20% of the time. I have a SharePoint Designer 2010 workflow that creates an item in a document library. The library has a Word template as the default which has field references that are populated upon item creation. The workflow fails, but only intermittently. The errors screenshotted here http://imgur.com/a/A73EO#2 appear in the ULS logs on each failure.
Examination of the workflow log history in the UI reveal two errors. The first shows "Error Occurred" with the log reading that the item may have been deleted or the user may not have access to the item. The second reads "Canceled" with the log showing a Coercion failure due to mismatched field types. There is no consistency as to which of the two errors described here occur. It can be one or the other.
We have verified all field types are correct. We also assume if there was a field type mismatch than the error would happen each time the workflow is ran, not intermittently. We also know that the item has not been deleted and that the user has access to it.
Originally, we hypothesized that this error was due to lack of resources on the servers. Server resource logs did not show an increase in usage and that the servers were minimally utilized (5-6%). This disproved the hypothesis and sent us back to the drawing board.
The workflow timer job is set to run every 5 minutes, with a batch size of 100 and throttle of 15.
I'm not sure what more information I can give and I am completely stumped. Any help is greatly appreciated.