RE: SolNET suffers a client bug

Magnus Hagander (mha@edu.sollentuna.se)
Sat, 24 May 1997 18:43:03 +0200


>Yesterday this announcement was posted to the SolNET mailing list:
>
Sure was :) Here is a little comment for _your_ list as well :)

<snip announcement>

>Since the announcement their keyrate has dropped from their recent
>2000 MKeys/sec to 738 Mkeys/sec. (We're about 3600 MKeys/sec). Their
>stats are worth a look - especially the big drop in the "blocks over
>last 24 hours" graph.
>
This mainly occured because all the old clients were prevented to
connect to the servers. And since it's a weekend.. Well. People just
don't work on weekends (except us, that is).

>
>
>It's worth taking a step back and learning from SolNET's recent
>mishap. Is anyone on this list actively following SolNET's
>development? What exactly was the bug? Can someone comment on their
>quality control versus ours?
>
I am not 1000% sure on _what_ the bug was, but the net result was that
only half the given keyspace was searched by the client.

>
>Did their keyrate have to go to 0 when they invalidated 1.11 clients.
>
Yes. Unfortunatly, only two clients or so were up and running within a
few seconds: Our own workstations :)

>
>What happened to all the (presumably non-broken) clients older than
>1.11? Surely not everyone had upgraded from versions before 1.11? Were
>they invalidated as well? Something to remember when designing
>protocols - make it possible to obsolete particular versions, not just
>"all versions older than X".
>
Due to the current design of our protocol, we had to invalidate _all_
clients. Both 1.11 and the working 1.02. This is naturally being looked
over into the next release... One should always be able to isolate a
non-working client version and just refuse that one. That was not
possible with the protocol we use.. :(

//Magnus
The SolNET DES Team

>
>