Please consider the following setup:
Internet -> |Ironport C350| -> |inner Mailgw| -> |Messagestore|
MX record are:
20 smtphost.ise.fraunhofer.de. <-- inner Mailgw
50 ipo1.ise.fraunhofer.de. <-- Ironport C350
(mailgw[1,2,b1] are backup gateways and .fhg is short for .fraunhofer)
The following problem occured when we switched off our inner mail gateway for maintenance: ipo1.ise.fhg.de received mails, could not reach smtphost and then routed them to mailgw1.fhg.de, which in turn handed them back to ipo1 and so on and so forth until the hop limit was reached.
Maybe I've misinterpreted the RFC2822 but I always thought that once a mailexchanger got a mail ii should never forward it to another mailexchanger with a lower priority (higher number).