Issue

• When device-mapper-multipath reloads it's configuration, sometimes the same scsi device path appears in more than one multipath device. When the incorrectly duplicated path refers to a different LUN, higher software layers (e.g. LVM, Oracle ASM, Filesystems) can be corrupted.
• If the incorrect LUN is the same size as the correct LUN then this can lead to silent corruption, which is particularly hard to diagnose.
• If the incorrect LUN is smaller than the correct LUN, the following syslog messages may be reported:

Dec 13 13:38:01 somehost multipathd: reconfigure (operator) 
Dec 13 13:38:01 somehost multipathd: mpath0: stop event checker thread 
Dec 13 13:38:01 somehost multipathd: mpath1: stop event checker thread 
Dec 13 13:38:01 somehost multipathd: HPTEST01: stop event checker thread 
Dec 13 13:38:01 somehost multipathd: mpath4: stop event checker thread 
Dec 13 13:38:01 somehost multipathd: mpath5: stop event checker thread 
Dec 13 13:38:01 somehost multipathd: HPTEST02: stop event checker thread 
Dec 13 13:38:01 somehost multipathd: mpath7: stop event checker thread 
Dec 13 13:38:01 somehost multipathd: HPTEST01: load table [0 2048000 multipath 0 0 8 1 round-robin 0 1 1 65:240 1000 round-robin 0 1 1 66:96 1000 round-robin 0 1 1  
Dec 13 13:38:01 somehost kernel: attempt to access beyond end of device
Dec 13 13:38:01 somehost kernel: dm-4: rw=17, want=11539463, limit=2048000