[aprssig] Rotate.aprs.net technical question (Was: APRS IS Issue ?)

[Phil Pacier]

Joel Maslak wrote:
> On Mar 2, 2008, at 3:41 PM, Rick Green wrote:
>
>> In the case of an error like this, does the client automatically go 
>> back and do another DNS query, (hopefully) get a different IP 
>> address, and retry the conection?  If so, how often, for how long?
>>   If not, what operator intervention is needed to force a retry?  
>> Restart the interface? Close and reopen the application? Force a 
>> flush of the workstation's DNS cache?  Force a flush of your router's 
>> DNS cache?  All of the above???
>
>
> My concern is...what if we lose one or two core servers for some 
> reason?  Is the redundancy adequate to handle the load?
>
> Last week, it did not appear like it was.  If one server was lost, 
> when the other was already refusing connections, the remaining server 
> was configured in a way which would have rejected stuff.
>
> So, my question: How many core servers must be operational to handle 
> the current number of connections (not CWOP load), today, of the APRS 
> core?  Do we have that? 

If I may be so bold, this is exactly why those of us at Tier 2 have been
promoting the tiered server concept for so long.  If the core were to
function as an interconnect hub for level 2 servers, the load and
redundancy could then be distributed among the level 2 servers.  Tier 2
currently has 37 servers in its network.  In fact, I have had to limit
the number of servers in the rotate address to a maximum of 22 because
certain routers will not parse more than that number of addresses!

73
Phil Pacier - AD6NH
www.aprs2.net