I have already logged a SR to EMC.
one of EMC appointed reseller, Unisys CE recommend to replace the server_3 h/w.
But from the error on celerra manager, i have my doubts whether its truly a h/w error.

If its not transient, please open a service request

there are many times when you see errors in Celerra Manager that are symptoms of a problem ..not necessarily the cause. It won't hurt anything to get it replaced since it's a standby datamover (in your configuration i assume). I had an instance where there was a bad memory module on one of the datamovers. EMC came out and replaced the whole datamover, they don't replace single parts.

Hi Dynamox
Is it an online type of replacement for the datamover.
Is it something like a hot-pluggable type of device where the faulty hw can
be replaced with a new hw without downtime.

yep, it's online. EMC came out onsite with a brand new datamover, i have an NS40 so this datamover looks kind of like a SPS battery on clariion, very thin and long. Once i failed over from primary to secondary, the CE un-screwed primary datamover, pulled it out and slid the new one back in. I ran "/nas/sbin/getreason" to see when it was back online and initiated a fail back.

Is there any risk? If lets say while replacing the faulty data mover online, and suddenly the working data mover suddenly goes bad, will the data be affected/corrupted?

nobody at EMC will say 100% that it will never happen but i find datamovers/Celerra to be very reliable, enterprise class platform. Do datamover panics occur ? you bet ..last year i only had one panic occur and i have 1 CFS14, 1 NS20, 1 NS80 and 2 NS40, standby datamover on CFS14 took over and we kept on chugging along.