On 01/06/2014 09:00 PM, Steve Dougherty wrote:
> TheSeeker points out a bug in negtype 10 that only affects seednodes.
> We're investigating and may have to release another update.

Our current understanding is that when a 1458 node attempts to connect
to a seed node with negtype 10 the seed node logs an error, and the
client seems to fall back to another negtype. This means on seed nodes
there is log spew of

(freenet.node.FNPPacketMangler, UdpSocketHandler for port XXXXXX,
ERROR): Unknown neg type: 10

but seeding still seems to function.

TheSeeker wrote a patch that avoids the message and presumably allows
seed nodes to use negtype 10 with clients. [0] Given that this does not
disrupt network function and only affects seed nodes, (and even then not
severely) it would make sense to me to wait until the next (hopefully
soon) release to include this fix.

In terms of the code that allowed this to happen I think it'd be good to
have an isValidNegtype() function to avoid duplication of such
conditions and add seed node tests. Are there longer-running seed node
tests already that aren't part of the standard test suite? Would it be
appropriate to run them when making a release?

Is the load on seed nodes after an update problematically high? Would it
help to have nodes wait a random interval after fetching an update
before restarting?

- Steve

[0]
https://github.com/freenet/fred-staging/commit/2faae2b00b8c56d2db8b157e8f95c8bb12f97303

Attachment: signature.asc
Description: OpenPGP digital signature

_______________________________________________
Devl mailing list
[email protected]
https://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl

Reply via email to