Over the last week of so my system becomes un-reachable either from the web-interface, network neighborhood from other machines and via telnet.  What I would typically do is a hard re-boot, wait for it to beep, go into the admin console as stop the new parity run and all was back until the next time this would happen.  Right now my Tower is in this state BUT there is a twist, my wife was playing some MP3's this morning in the kitchen and they are still playing fine, yet I can't reach the machine at all via telnet or the web admin screen, I can however ping the box.

 

This system has been VERY stable since it was upgraded to 3.0 until I had that funky disk6 problem that would cause a parity check to lock-up the system.  That was corrected by replacing disk6 with a new drive and all seemed normal, but I guess it's not.  Any ideas...?

 

UPDATE - I have noticed one other strange and new event.  I have a webserver on my network that is running a software firewall program.  For the last several days the unraid has been logging events in this software that I have never seen it do before.  The event is not considered a threat so the software allows it but does log it, the event is called "TCP_Port_Scan" and it has never done this before. The event is described as "This signature event detects when a computer attempts to access more than 10 different TCP ports on a protected computer" 

I have no idea what TCP_Port_Scan means..., sorry.  One thing you might check is the lease times generated by your DHCP server.  Could be that your server's IP address lease expires & it tries to get a new lease and can't get one.  You might try looking at the system log via the console (keyboard & monitor) to see if there are some DHCP releated messages in there.

I have done a little more research and it's clear that the unraid never did these port_scan events before, it started about a week ago when I started having parity issues.  I no longer have the parity issue but the box does go unreachable a couple of times a day.  It also looks like when the box goes unreachable is right after a large TCP_Port_Scan event takes place.

 

DHCP is set to never let leases expire so that's not it.