# how are some nomadnet hosts announcing every minute? _Help · started by NONGKHAEM_BKK_TH on Fri, Sep 18, 2026 9:17 AM_ --- ## Original post **NONGKHAEM_BKK_TH** · Fri, Sep 18, 2026 9:17 AM Hello, I'm knee deep in the rns docs, and I can't figure out why some nomadnet nodes (and probably other types) are announcing at rates far higher I thought the network would have allowed. In particular, using rbrowser to count the announces. VarnaTransport is announcing every minute or so, 852 announces over 14h hours or so. Shadow: 170 (or 12 times an hour). RNSfaultline: 170. cincinnatus: 171. You get the point - I know that announce_interval can be set to whatever, but I thought announce_rate_target was here to prevent this behaviour, and had a default of 3600. I understand that that if I was a TCP client to an interface set to announce_rate_target == 0 AND was at the same time 1 hop away from the offending nomadnet announcer, this could happen. But VarnaTransport is 128 hops away (so basically go figure this out), Shadow is 3 hops away, cincinnatus 2. None of these are 1 hop away. I guess I'm demonstrating my total lack of understanding of how this works exactly, and I suppose there is a non-zero chance that there are enough transport nodes preferring to rely on the 2% bandwidth rule to allow these announces to make their way to my machine, but I still find it puzzling. Can anyone shed some light on this? Thank you so much. --- ## Reply 1 **aetherlab** · Fri, Sep 18, 2026 9:28 AM It's simple - poorly configured. I have no idea why someone would consider this a good thing at all. 1 hour is ample, 6 hours is more reasonable. Even less. --- ## Reply 2 **K8 - Mobile** · Fri, Sep 18, 2026 10:33 AM I'm not at my computer to investigate more, but you can run `rnpath -r` to get specific info about announce rates for destinations. For transport nodes, `announce_rate_target` does default to 3600 with a grace of 5. I've seen the rate limiting working on my node. I'm not sure why a public interface would go out of their way to disable the rate limiting unless they were trying to gather announce statistics or something, or if it was running a broken implementation. Also, a hop count of 2 means that both you and the destination are connected to the same transport node. --- ## Reply 3 **NONGKHAEM_BKK_TH** · Fri, Sep 18, 2026 10:51 AM Thank you aetherlab - I should have added - this already has consequences. rBrowser has a built-in delay to start auto-indexing further pages as part of its crawler, which can never trigger because the constant announces reset it to zero. I've reported this to the creator of rBrowser, but it makes me wonder - isn't the network defense against such 'spam' the very rate_limit and bandwidth limit settings so many transports seems to have set to silly values? My understanding is as long as enough transport nodes form a path to any given machine, these announces will make it through, leaving it to the app layer to maybe auto-blackhole flood sources, but ... clearly that's something that should be handled at protocol level instead (AFAIK no client implements this)? Maybe I misunderstand or overthink this :) --- ## Reply 4 **circfruit** · Fri, Sep 18, 2026 10:51 AM Oh, shit! VarnaTransport is mine. Obviously I did some testing and forgot to lower the announce rate 🙄 I'll reduce it to a sensible one. Apologies if this has caused troubles. --- ## Reply 5 **circfruit** · Fri, Sep 18, 2026 11:07 AM Also to be specific, the Nomadnet node is running on my backbone. The latter was running rns 1.5.0 and no specific configuration than default. Could it be the case that 1.5.0 have a bug? Anyway I've upgraded to 1.5.5 now and reduced the offending nomadnet node announce to 1.5h. --- ## Reply 6 **Mark** · Fri, Sep 18, 2026 4:39 PM Lol @circfruit, that's what happens. If there's an annouce rate target configured on your transport node, but the nomadnet instance is running *on that same node*, using the shared instance, it will bypass the announce rate target. It only applies for announces that come *in*, and would be rebroadcasted onto the network, so all the clients connected to your transport would get the constant announces. So for example @NONGKHAEM_BKK_TH, if you were connected directly to @circfruits transport from your rBrowser device, you would see all the announces coming in from the node, since no other transport nodes along the way would be able to catch them in their announce rate target filtering. --- ## Reply 7 **circfruit** · Fri, Sep 18, 2026 5:49 PM **Mark** wrote: > If there's an annouce rate target configured on your transport node, but the nomadnet instance is running on that same node, using the shared instance, it will bypass the announce rate target. It only applies for announces that come in, and would be rebroadcasted onto the network, so all the clients connected to your transport would get the constant announces. Oh, that makes sense now. Thanks. Indeed Nomadnet node is using the shared rnsd with the backbone interface. ---