I came across a scenario where on our staging and production environments a service call was failing, while it would succeed in our development and QA environments, all using the same data. It seemed to be caused by a sufficiently large response, but only on some of the servers.
By setting dataContractSerializer maxItemsInObjectGraph to a high enough value in the web.config we were able to overcome the issue, however I would like to understand why it was only failing on some servers, rather than all. (There was also no error logged on failure, it simply returned an empty response, so not much to go on there)
While our staging/qa/production environments are all supposed to be the same, in practice they have small differences, and I'm wondering if some setting on IIS is possibly overriding a default on some of the servers.
Can anyone offer any insight?