On Fri, Oct 18, 2002 at 01:31:23AM -0400, Michael Wiktowy wrote: > > > > > >From: > >Ian Clarke <ian at freenetproject.org> > >Date: > >Thu, 17 Oct 2002 09:49:26 -0700 > > > > > >I am finding that it can take as much as 3 minutes to load the routing > >table. > > > >This could be problematic since many users will have given up after such > >a long delay deciding that Freenet wasn't working. > > > >The store directory contains 3MB of files, a total of 465 individual > >files as far as I can tell. A normal-sized store is likely to be > >10 or 20 times this size. > > > >Is there any way we can reduce this huge delay? > > > > > > > >Ian. > > > > > > (once again I apologise for the broken threading due to digest replying) > > I am finding similar delays with a default size datastore/routing table. > It can be a small as 1 minute but typically is about 3 minutes and mine > is a full 200MB wit 2141 keys so it seems appearently independant of > store-size. Is it in loading routing table, or is it in loading temp bucket factory? If the latter, then someone or something is on crack, probably the JVM, and I'm going to try to reproduce it. My 1GB node, with logLevel=debug, takes about 15 minutes to load with another one running in the background... and almost all of that is in "loading temp bucket factory". Ok, you've convinced me that there is a problem, I'll look into it. > > One suggestion that I have, if there is nothing that can be done to shorten > loading time (hey lets make a monolithic datastore that is super efficient > *runs away from the lynch-mob ;]*), is to have mainport return a "freenet > is loading ... please wait" page that will autorefresh and load the > requested > URI when the node is all done whatever it does that takes so long on > startup. That would at least make it a bit more user-friendly. Is there a > problem with starting mainport before the node and then having it wait > for a proper FCP to open up. > > Another avenue that might be followed is modifying the shutdown scripts > to somehow checkpoint freenet so that it does not have to rebuild its brain > on startup each time. I make this suggestion with no understanding of > what the node is doing on startup that is so processor intensive. So pardon > my ignorance if this is silly. I've always just wished that the stop-freenet > script did a little more then drop-kicking the java processes and pulling > the rug out from under all the connections to the node. > > Mike > > > _______________________________________________ > devl mailing list > devl at freenetproject.org > http://hawk.freenetproject.org/cgi-bin/mailman/listinfo/devl >
-- Matthew Toseland toad at amphibian.dyndns.org amphibian at users.sourceforge.net Freenet/Coldstore open source hacker. Employed full time by Freenet Project Inc. from 11/9/02 to 11/11/02. http://freenetproject.org/ -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 189 bytes Desc: not available URL: <https://emu.freenetproject.org/pipermail/devl/attachments/20021018/a09325e4/attachment.pgp>
