On the Acme Packet 4600, certain unexpected power events can cause the Ethernet ports to enter a 'NotReady' state. If this happens and you have not enabled fault-recovery rebooting, you must manually power cycle the system to restore normal functionality. Because the Ethernet ports are down, you cannot power cycle over the network.

You can identify this problem by finding PCIe link lost in the kernel.log file.

Hotswap event: PHY0 Hotswap status reg: 0x05
Hotswap event: PHY0 not ready
Hotswap event: PHY0 latch closed
Hotswap event: PHY0 present
pcieport 0000:02:08.0: pciehp: Slot(72): Link Down
pcieport 0000:02:08.0: pciehp: Slot(72): Already disabled
pcieport 0000:02:08.0: pciehp: Slot(72): Card present
pcieport 0000:02:08.0: pciehp: Slot(72): Latch open
igb 0000:0d:00.0 wancom0: PCIe link lost
igb 0000:0d:00.2 wancom2: PCIe link lost
igb 0000:0d:00.1 wancom1: PCIe link lost
bmr464 i2c-34/34-0036 POL fault in status reg 0x79: 0x8860
bmr464 i2c-34/34-0036 POL fault in status reg 0x7a: 0x0080
igb 0000:0d:00.3 eth3: PCIe link lost

If you've encountered this problem, Oracle recommends enabling fault-recovery rebooting to automatically restore your system. With this enabled, the Oracle Communications Session Border Controller gracefully shuts down and restarts when it detects all the Ethernet ports have gone down. To enable automatic fault recovery, navigate to the system-config element and set the fault-recovery attribute to enabled.

ORACLE# conf term
ORACLE(configure)# system
ORACLE(system)# system-config    
ORACLE(system-config)# select
ORACLE(system-config)# fault-recovery enabled 
ORACLE(system-config)# done

By default this parameter is disabled.