There have been reported incidences of reed relays (but it could equally have been an EMR) being adversely effected by hard disc drives in Slot 1 of a PXI chassis.
The PXI standard makes no claims about compatibility of modules when placed in a chassis, it is largely left to the vendors to do the right things to ensure that modules are made to be compatible and to resolve the issues on a case by case basis. This makes it easier and faster to develop products since there are no complex specifications to comply with (as for example there is in VXI).
Hard Disc Drive Issues
Slot 1 controllers usually include a hard disc drive and hard disc drives contain very strong magnetics. Relays rely on magnetic fields to close mechanical contacts, and rely on a spring pressure to release the contact when a field is applied. The open and release magnetic field strengths are rarely the same because of the presence of magnetic materials in the construction that result in the closing field being higher than opening field. Whether it is a reed relay or an EMR this is usually seen as a difference in the pull in voltage applied to compared to the release voltage.
In the reported example a relay module was located in Slot 2 and an embedded controller in Slot 1 which had a relatively badly shielded magnetic system that was close to the the right hand edge of the module. A relay would "fail" because a contact would close but then would not open when required but separate tests indicated no fault with the relay. The only way to reset the relay was to switch off the chassis and remove the module so that the relay would released. The relay in the module was magnetically screened and had no issues when operated with other relays on the module which were much closer than the hard drive. The relays had significantly better magnetic screening than many other reed relay designs.
The problem appears to be only associated with a small proportion of embedded controller fitted with this particular hard drive - only one incidence has been recorded.
The solution was to simply move the switching module to Slot 3 or higher.