On October 01, 2003 07:57 am, Toad wrote: > On Wed, Oct 01, 2003 at 07:44:35AM -0400, Ed Tomlinson wrote: > > Hi, > > > > I have noticed something funny in the ocm stats. My box with a node > > up 14 hours (6214) has 242 connections transmitting data and has > > 324M of data queued on them. My output bandwidth is limited to > > 10k/s. The servlet/nodeinfo/performance/general tells me I am > > using about 8k/s. So either the ocm lies about what is queued OR > > most of those connections are stalled... It may well be a stalling > > problem - if I limit the send queue size to a couple of mins of > > traffic (by QRing) servlet/nodeinfo/performance/general shows > > dropping bandwidth over time. > > Hrrm. Well, they could be limited by the inbound on the dest nodes, or by > the outbound on the source node (unless they are reading from the store > - it'd be interesting to know how many were reading from store and how > many were straight-through).
I would suspect that most were reading from store as the input queue was about 11M at this time. This would tend to imply that the transmissions were stalled... > > Are we timing out transmits? I think I will _experiment_ with something that looks for connections that have not seen any progress in fixed window (say 5 mins). > I don't think so, but TCP might be. One more reason for multiplexing. > > > Another interesting point. We I coded the first NG implementation > > I ignored Transfer Failed messages. I put some metrics in to check > > for them and there were so few that it was safe to ignore them (its > > not safe in the long run). Now we are seeing lots of transfer failed. > > Why? > > You may not have counted all possible sources of transferFailed at that > point. Please find the *proximate* cause - many may be caused by, for > example, the receiving connection failing, or getting restarted. Also > since that is nonfatal, the use of it in the formula may be wrong. That is possible - just hooked into the transferFailed method. If it was not called correctly and this as since been fixed then this is the case. > > Think answering any of these questions will solve the others. > > > > Anyone have ideas? > > Ed _______________________________________________ Devl mailing list [EMAIL PROTECTED] http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/devl
