Many network protocols rely on temporal partitions to provide isolation between different nodes (TDMA slots) or different traffic classes (multi-phase cyclic frameworks). Typically, the duration of the slots or phases is not correlated with the duration of packet transmissions, which is variable and non-preemptive. Thus, it is possible that the limit of the slot or phase be overrun by an ongoing packet transmission or, if this cannot be tolerated, idletime must be inserted at the end of the slot or phase whenever a packet does not fit in. Nevertheless, both situations lead to scheduling anomalies in which the worst-case network delay does not occur necessarily with the synchronous release of all other packets, or just the higher priority ones. This paper highlights two such anomalies showing their origin and indicating that, in such circumstances, it is not possible to determine the worst-case network delay with exactitude in the general case. However, it is still possible to upper bound the network delay and the paper shows non-optimal solutions for those cases.