Timestamps are in GMT/BST.
[0:01] * zerwas (i=zerwas@) has joined #freenet
[0:16] * Queridinhozinho está away; Auto-away; inativo por 1801 segs - site[n/a] email[none@none] uin[none] log[on] pager[off] -=[Cyber]=-
[0:21] * Queridinhozinho (n=queridin@) Quit ("CyberScript - Salve as tartaruguinhas marinha. Use CyberScript. (www.cyberscript.org)")
[0:25] * Dieppe (n=clement@) Quit (Remote closed the connection)
[1:00] * Tarantuka (n=eddie207@) Quit ("Quitte")
[1:11] * toad_ (n=toad@) Quit (Remote closed the connection)
[1:13] * phrosty (i=phrosty@) has joined #freenet
[1:24] * sbc (n=ca@) Quit ("Ex-Chat")
[1:38] * NEOatNHNG (n=michael@) has joined #freenet
[1:39] * NEOatNHNG1 (n=michael@) Quit (Read error: 110 (Connection timed out))
[2:05] * ghoul_ (n=ghoul_@) has joined #freenet
[2:09] * zerwas_ (i=zerwas@) has joined #freenet
[2:10] * zerwas (i=zerwas@) Quit (Nick collision from services.)
[2:10] * zerwas_ is now known as zerwas
[2:10] * ghoul_ (n=ghoul_@) Quit ("ChatZilla 0.9.83 [Firefox 3.0/2008052906]")
[2:16] * gasi_ (n=gasi@) has joined #freenet
[2:24] * archangelpetro (n=hellequi@) Quit ("Build a hundred bridges and suck a single cock, and you're not a bridge builder.. you're a cocksucker")
[2:31] * robink (n=robink@) has joined #freenet
[2:33] * gasi__ (n=gasi@) Quit (Read error: 110 (Connection timed out))
[2:37] <robink> Is insecure mode the same as promiscuous mode?
[2:42] * caytchen (n=caytchen@) Quit ("Verlassend")
[2:45] * [Nande] (n=KVIrc@) has joined #freenet
[2:47] * Nande (n=KVIrc@) Quit (Read error: 60 (Operation timed out))
[3:23] * Azerthoth (n=Azerthot@) has joined #freenet
[3:34] <TheSeeker> yes
[3:55] * crudpuppy (n=someone@) has joined #freenet
[4:06] <crudpuppy> anyone alive?
[4:07] <crudpuppy> I got the linux install complete but seems it will not fully start
[4:22] * v1rtuous (i=centr1c@) Quit ("Ex-Chat")
[4:23] * rtrfix (i=Xulfa@) has joined #freenet
[4:39] * n0ob (n=travis@) has joined #freenet
[4:45] * Azerthoth (n=Azerthot@) has left #freenet
[5:00] <mrdmx-away> crudpuppy: what java are you using?
[5:00] * mrdmx-away is now known as mrdmx
[5:06] <crudpuppy> 1.6 something
[5:07] <mrdmx> install sun java
[5:07] <crudpuppy> it is sun jre
[5:08] <mrdmx> hmm
[5:08] <crudpuppy> java version "1.6.0_05"
[5:08] <mrdmx> what exactly does not start?
[5:09] <crudpuppy> this is a gentoo system by the by if that matters
[5:09] <crudpuppy> last line : Optimized native BigInteger library 'net/i2p/util/libjbigi-linux-pentium4.so' loaded from resource
[5:09] <crudpuppy> that I get in the browser
[5:10] <crudpuppy> and it refreshs forever with no status change
[5:10] <mrdmx> :/
[5:11] <crudpuppy> actually
[5:11] <crudpuppy> it seems to be retrying
[5:11] <crudpuppy> every 10 minutes or so after stopping
[5:11] <crudpuppy> it restarts
[5:11] <crudpuppy> I just didnt notice it during refreshs
[5:11] <crudpuppy> says "Startup failed: Timed out waiting for signal from JVM."
[5:17] <mrdmx> any errors in wrapper.log?
[5:17] <crudpuppy> http://code.bulix.org/bhd2ls-67600
[5:18] <crudpuppy> this is the console output when started with console command
[5:18] <crudpuppy> that is pretty much as far as it gets before retrying
[5:20] <crudpuppy> I'm rebooting...lookin at wrapper.log it looks like it did start the first time and I musta killed out of ti
[5:20] <crudpuppy> and now it wont start
[5:20] <crudpuppy> so hopfully restart will clear up something in there
[5:20] <crudpuppy> hehe
[5:20] * crudpuppy (n=someone@) Quit ()
[5:25] <robink> I'm having a hard time finding peers to connect to, even in insecure mode.
[5:26] <mrdmx> weird
[5:26] <mrdmx> how many peers do you have?
[5:26] <robink> mrdmx: I have a few, but they're all offline.
[5:27] <mrdmx> all of them?
[5:27] <robink> mrdmx: Yeah.
[5:27] <mrdmx> :|
[5:28] <robink> I was hoping Freenet would rely on some seeded refs if I popped it into insecure mode.
[5:28] <mrdmx> yeah it should
[5:28] <robink> blrgh
[5:32] * [Nande] (n=KVIrc@) Quit ("byes, tengan la bondad de ser felices. ;)")
[5:33] <robink> There's only one key I want to pull.
[5:34] <mrdmx> did you restart the node?
[5:34] <mrdmx> im not sure if you have to
[5:35] <robink> Yes.
[5:35] <mrdmx> hmm
[5:36] <mrdmx> try downloading the latest seednodes :http://mirror3.freenetproject.org//alpha/opennet/seednodes.fref
[5:37] * Luke771 (n=luke@) has joined #freenet
[5:46] <robink> OK, added all non-testnet refs but am unable to connect to them.
[5:49] <Luke771> say again?
[5:49] <Luke771> what have you done exactly?
[5:50] <Luke771> what are non-testnet refs?
[5:50] * Luke771 has been using freenet for a couple of years and has no idea what non-testnet refs are
[5:50] <robink> I fed the non-testnet refs in that URL to my Freenode node, and it recognises the refs but won't connect to the hosts they are associated with.
[5:50] <Luke771> ok
[5:50] <robink> My freenet node only accepts refs if testnet=false.
[5:50] <Luke771> delete them
[5:50] <robink> Why?
[5:50] <Luke771> sure
[5:51] <robink> What?
[5:51] <Luke771> hmm... because I say so?
[5:51] <robink> OK
[5:51] <Luke771> enable opennet
[5:51] <robink> I did.
[5:51] <Luke771> well then, check out your stranger page
[5:51] <Luke771> strangers
[5:51] <robink> No connected hosts.
[5:51] <Luke771> do you have any connections there?
[5:51] <robink> I was adding refs to the stranger page, not the friend page.
[5:51] <robink> Nope
[5:52] <Luke771> you don't have to add connections to the strangers page, that should be automagical
[5:52] <Luke771> ok, update the node
[5:52] <robink> Update the node?
[5:52] <Luke771> what OS are you running? windows?
[5:52] <robink> Linux.
[5:52] <Luke771> ok
[5:52] <Luke771> then cd to the freenet dir and do ./update.sh
[5:53] <robink> I'm told it to update via the web interface
[5:53] <robink> I'll see if that works first.
[5:53] <Luke771> you may need to chmod +x update.sh
[5:53] <Luke771> your problem is probably about the seednodes.fref file
[5:53] <robink> ok
[5:53] <Luke771> and you're unlucky: I don't know much about that file
[5:54] <Luke771> you can try to get a newer seednodes.fdref from downloads.freenetproject.org
[5:54] <robink> hm
[5:54] <Luke771> my idea was that updating the node after enabling opennet _maybe_ you'll get a newer seednodes.fref
[5:55] <Luke771> but I don't really know
[5:55] <robink> update.sh failed.
[5:56] <Luke771> also, when you add refs to the friends page (or to the strangers page for that matter) you can't simply add references, you and the node you add have to add each other, or it won't work
[5:56] <Luke771> ok, get a newer update.sh from downloads.freenetproject.org
[5:56] <robink> I thought if it's a promiscuous node and the ref I'm adding is of another open peer, then that shouldn't be an issue.
[5:56] <Luke771> btw, how do you installed? you used the automagical installer or the tarball?
[5:57] <Luke771> er "hod did you install", I mean
[5:57] <Luke771> w
[5:57] <robink> I used an ebuild.
[5:57] <Luke771> hm
[5:58] <robink> Essentially an automagical install.
[5:58] <Luke771> never done that (in fact, I'm not even sure what that is)
[5:58] <Luke771> yeah but
[5:58] <Luke771> you know what
[5:58] <Luke771> ok, try as I siaid first, that might work
[5:58] <Luke771> get a new update.sh and run it
[6:00] <Luke771> if it doesn't work with an update.sh you just got from the website, the most quick way to make it work is to delete your whole freenet dir and run the autoinstaller directly off the download page, clicking on the "download freenet" link
[6:00] <robink> 'k
[6:00] <robink> I'd install a much newer version using an ebuild, but there's a circular dep issue.
[6:01] <Luke771> that would be, not downloads.freenetproject.org but the frontpage, just freenetproject.org: click on download freenet, and then on a link in the middle of the text blah blah blah _download Freenet_ blah blah blah
[6:01] <Luke771> why can't you simply use the new_installer.jar ?
[6:02] <Luke771> (the file that is executed when you click on that link)
[6:02] <robink> Updated freenet with a newer update.sh
[6:02] <robink> I dunno, it just fails silently.
[6:02] <Luke771> or, you could use the tarball freenet.tar.gz, extract it to a freenet dir, and run update.sh
[6:02] <Luke771> it failed again?
[6:03] <Luke771> what fails silently, the new_installer.jar?
[6:04] <robink> No, just that using the new_installer.jar failed silently when I tried it
[6:04] <robink> Using a new update.sh works quite well
[6:04] <Luke771> what java you got?
[6:04] <robink> 1.6.0_03
[6:04] <Luke771> sun?
[6:04] <robink> Yeah
[6:04] <Luke771> hm
[6:05] <Luke771> well, the update.sh did work this time, right?
[6:05] <robink> Yes.
[6:05] <Luke771> ok
[6:05] <robink> Nothing yet, I downloaded a new seednodes.fref file and placed it in the freenet directory.
[6:05] <Luke771> lets see if the opennet connects now
[6:05] <robink> The node just slowed waaay down, hopefully it's busy trying to connect to good nodes.
[6:05] <Luke771> not even a couple of not-yet-connected nodes?
[6:06] <robink> There's the ones I added, haven't checked real recently.
[6:06] <Luke771> in the strangers page
[6:06] <Luke771> take them away, they won't connect
[6:06] <robink> Gimmie a sec, like I said the node is slooooooowww.
[6:06] <robink> I can't remove them.
[6:06] <Luke771> hm
[6:06] <Luke771> stop the node
[6:07] <Luke771> edit peers-open-(portnr.) or something like that
[6:07] <Luke771> delete everuthing from it and leave it blank
[6:07] <Luke771> restart the node
[6:07] <robink> OK, restarted the node
[6:08] <Luke771> now the nodes you added should be gone
[6:08] <robink> Only the old noderefs are showing.
[6:08] <Luke771> are they
[6:08] <robink> Nope
[6:08] <Luke771> hmpf
[6:08] <Luke771> you sure you edited the right file?
[6:08] <Luke771> checking out filename
[6:08] <robink> It's trying new nodes
[6:09] <Luke771> ok, then it works
[6:09] <robink> Wait, it just tried one new node
[6:10] <robink> Oh wait, there it goes again
[6:10] * christefano (n=christef@) Quit ("time for a coffee")
[6:10] <Luke771> but the problem is how to get rid of those you added, they won't connect anyway... I _hope_ the node will dump them automatically when they don't connect after a while but I'm not sure it will
[6:10] <robink> It's probably going slower do to the fact that there are already a bunch of noderefs on the stranger config page, but it's going.
[6:10] <Luke771> ok
[6:10] <Luke771> stop the node with ./run.sh stop
[6:11] <robink> Connected!
[6:11] <Luke771> browse to your freenet dir
[6:11] <robink> Why should I stop the node?
[6:11] <Luke771> and delete peers-open-<port#>
[6:11] <Luke771> hmm... because I say so?
[6:11] <robink> Even though it's connected?
[6:11] <Luke771> to delete the peers that you added, they won't connect anyway
[6:12] <Luke771> no problem it will get new peers
[6:12] <robink> ok, letting it recover a little
[6:12] <Luke771> I'm not sure that the node will get rid of them automatically, because you added them manually
[6:13] <Luke771> anyhow, except for those manually added nodes, you're pretty much OK now
[6:13] <Luke771> also when you add nodes to the darknet (friends page) make sure that you add each other, one-way adding won't work
[6:36] <robink> Luke771: Gotcha
[6:38] <Luke771> what?
[6:38] <Luke771> was that a joke?
[6:38] <robink> Luke771: Uh, no.
[6:39] <Luke771> :P
[6:39] <Luke771> is it OK now?
[6:39] <robink> Yeah, it's doing great.
[6:39] <Luke771> cool
[6:39] <robink> Upgrading the client and getting a new seednode.refs file fixed everything.
[6:46] * zerwas (i=zerwas@) Quit ("Keep on flowing ...")
[7:01] * Pichu0102 (n=pichu010@) Quit (Remote closed the connection)
[7:01] * n0ob (n=travis@) Quit ("Now onto something more interesting")
[7:01] * Pichu0102 (n=pichu010@) has joined #freenet
[7:09] * zerwas (i=zerwas@) has joined #freenet
[7:10] * zerwas (i=zerwas@) Quit (Client Quit)
[7:19] * dbkr- (n=nnndbkr@) has joined #freenet
[7:35] * zerwas (i=zerwas@) has joined #freenet
[7:36] * dbkr (n=nndbkr@) Quit (Read error: 110 (Connection timed out))
[7:37] * dbkr- is now known as dbkr
[7:37] * zerwas (i=zerwas@) Quit (Client Quit)
[7:52] <nextgens> hi
[7:56] <Luke771> hi
[8:05] * Ratchet (n=xxxx@) has joined #freenet
[8:17] <phrosty> nextgens: did my email ever hit the devl list
[8:22] <nextgens> which one?
[8:23] <phrosty> sent yesterday, about scalability
[8:23] <nextgens> yes it did
[8:23] <nextgens> and ian replied iirc
[8:24] <phrosty> ahh i see, i wonder how that slipped by me
[9:48] <FreenetLogBot> r21156 (1153) was built successfully on emu, mirrors are updating
[10:30] * FrinkC (n=FrinkC@) has joined #freenet
[10:54] * toad_ (n=toad@) has joined #freenet
[10:54] * ChanServ sets mode +o toad_
[11:07] * sbc (n=ca@) has joined #freenet
[11:24] * ahuxley (n=ahuxley@) Quit ()
[11:46] * infinity0 (n=infinity@) has joined #freenet
[11:49] * infinity0 (n=infinity@) Quit (Read error: 104 (Connection reset by peer))
[11:49] * infinity0 (n=infinity@) has joined #freenet
[11:50] <infinity0> where do i report bugs?
[11:51] * infinity0 (n=infinity@) Quit (Read error: 104 (Connection reset by peer))
[11:52] * infinity0 (n=infinity@) has joined #freenet
[11:52] <sbc> infinity0: https://bugs.freenetproject.org
[11:52] <infinity0> ah, thanks
[11:56] * infinity0 (n=infinity@) Quit (Read error: 104 (Connection reset by peer))
[11:56] * infinity0 (n=infinity@) has joined #freenet
[11:58] * zerwas (i=zerwas@) has joined #freenet
[11:58] * DaffyDuck_ (n=jan@) Quit ("Leaving")
[12:03] <nextgens> infinity0> what's the problem?
[12:03] <infinity0> basically, http://wiki.freenetproject.org/Freenet0Point7withFreeVm is wrong
[12:04] <infinity0> freenet doesn't install through gcj, because gcj hasn't implemented the JKS algorithm
[12:04] <infinity0> but if you use the openjdk jvm, the install/update scripts work
[12:04] <nextgens> well, it's a wiki; update the page :)
[12:04] <infinity0> yeah i'm just not sure what to write though
[12:04] <infinity0> becaues it's like "developers officially support only gcj"
[12:04] <FreenetLogBot> r21160 (1153) was built successfully on emu, mirrors are updating
[12:05] <nextgens> infinity0> we have never really supported gcj
[12:05] <nextgens> we accept patches to workaround gcj bugs
[12:05] <nextgens> but that's about it
[12:06] <nextgens> the installer has never worked with gcj and there is no plan to fix it :/
[12:06] <infinity0> oh, ok, but that's a shame, i think all that's needed is to use PKCS12 keys instead of JKS ones
[12:07] <infinity0> but what should i write for the wiki? should i just delete the GIJ part?
[12:07] <nextgens> does sun's jvm support that?
[12:07] <infinity0> i dunno
[12:08] <nextgens> I'll have a look
[12:14] <Luke771> uninstaller.jar decides what l10n to use based on the IP?!?
[12:15] * sanity (n=ian@) has joined #freenet
[12:15] * ChanServ sets mode +o sanity
[12:15] <Luke771> I'm having problems with one of my testing nodes so I decided to re-do everything from scratch.. I thought once I was there I could try and run the uninstaller, just to see what it would do
[12:15] <Luke771> so I went to <freenet dir>/Uninstaller and ran java -jar uninstalle.jar
[12:16] <Luke771> the interface popped up in italian, despite the OS being in English and the Freenet l10n set to English
[12:16] <nextgens> http://java.sun.com/javase/6/docs/technotes/guides/security/jsse/JSSERefGuide.html
[12:16] <nextgens> infinity0> it doesn't support trustedCertEntries in PKCS12 keystores
[12:16] * archangelpetro (n=hellequi@) has joined #freenet
[12:17] <nextgens> Luke771> we can't assume that the node's setup isn't f*cked up in the uninstaller
[12:17] <nextgens> hence we don't read the l10n preferences from it
[12:17] <Luke771> sure
[12:17] <nextgens> and it's hardcoded at install time
[12:17] <nextgens> anyway, that's a really minor issue :)
[12:18] <Luke771> well at least use the language from the OS...
[12:18] <Luke771> yes totally a minor issue
[12:18] <Luke771> not even worth filing a bug
[12:18] <infinity0> oh, ok
[12:19] <nextgens> infinity0> anyway, if you've any suggestion...
[12:20] <nextgens> I don't like JKS either
[12:20] <nextgens> but as far as I know there is no way to have the same code for the different jvm vendors
[12:20] <infinity0> sorry, i don't have any ideas, i'm compeltely new to this
[12:20] <nextgens> and our priority atm is to get it working on sun's jvm
[12:20] <infinity0> was just trying to get freenet working on my debian box and read through this stuff about JKS and GCJ
[12:20] <infinity0> oh right ok
[12:21] <infinity0> well, it works on the openJDK JVM
[12:22] <infinity0> so what about the wiki page? i haven't tried the other JVMs but GCJ definitely can't install/update, even though it can run freenet perfectly fine
[12:23] <nextgens> put a note on the page saying that the install doesn't work :)
[12:23] <nextgens> I'll have a closer look to the gcj's way of doing things
[12:24] <infinity0> kk
[12:26] <toad_> infinity0: surely it can update? that's a separate question, it doesn't use the uninstaller
[12:27] <nextgens> toad_> no, since Sha1Test uses SSL now
[12:27] <sanity> # networkSizeEstimateSession: 4460 nodes
[12:27] <sanity> # networkSizeEstimate24h: 3869 nodes
[12:27] <sanity> not bad
[12:28] <sanity> toad_: remind me how those are determined?
[12:28] <nextgens> and SSL requires a keystore format gcj doesn't handle
[12:28] <toad_> ah, the update script doesn't work on gcj, i see
[12:28] <toad_> sanity: the number of locations seen in swapping
[12:28] <toad_> over X period
[12:28] <sanity> toad_: so it is probably an overestimate given that nodes can change locations?
[12:28] <nextgens> I'm gonna need some people to update to current trunk to help me try the FOAF-routing stuffs I've just commited
[12:29] <sanity> toad_: how goes db4o?
[12:29] <nextgens> sanity> not necessarily as swaps are seen on a split horizon
[12:30] <toad_> sanity: it would only be a major overestimate if we reset locations too often
[12:30] <nextgens> well, it depends on the topology... but it's likely to be an overestimate, yes
[12:30] <toad_> but iirc we reset locations every 2 days or so, so it might be 50% over
[12:30] <infinity0> toad_: it doesn't, gij fails with a "NoAlgorithmException" due to JKS
[12:30] <toad_> sanity: still debugging...
[12:30] * infinity0 (n=infinity@) Quit (Remote closed the connection)
[12:31] <toad_> sanity: somebody came up yesterday and said he was trying to get his company to use freenet for a sort of VPN, and inserts weren't working on his micro-darknet (9 nodes) ... should i spend time debugging this? if small darknets don't work then they will never get connected to other small darknets
[12:31] <sanity> toad_: it would be cool if someone with a long-running node uploaded that stat to the website once every day or so, perhaps even feeding it into a graph...
[12:31] * infinity0 (n=infinity@) has joined #freenet
[12:33] <toad_> sanity: agreed
[12:34] <toad_> nextgens: can we get CIA to subscribe to #freenet-testers instead of #freenet-chat ?
[12:34] <sanity> toad_: yes, that sounds like a worrying problem
[12:34] <sanity> toad_: could the insert just be RNFing?
[12:34] * infinity0 (n=infinity@) Quit (Read error: 54 (Connection reset by peer))
[12:35] <nextgens> toad_> sure
[12:35] <pmpp> big inserts on small darknets has something to do with garbage collecting : when netspeed is too high node run out of memory fast
[12:35] <nextgens> http://www.gnu.org/software/gnu-crypto/#keyring
[12:35] <nextgens> apparently gcj doesn't have any keyring
[12:35] <nextgens> can you belive that?
[12:36] <toad_> sanity: yeah but i thought we gave up when we reached 2 RNFs ... also, he said it worked for small stuff ...
[12:36] <toad_> < 200K
[12:36] <toad_> pmpp: he had it running out of threads
[12:36] * sanity_ (n=ian@) has joined #freenet
[12:36] * ChanServ sets mode +o sanity_
[12:37] <pmpp> toad_> maybe he tried many concurrents inserts ...
[12:38] <pmpp> anyway on a gigabit link he will run out of memory in most cases
[12:39] * infinity0 (n=infinity@) has joined #freenet
[12:39] <infinity0> sorry, pidgn keeps crashing
[12:39] <infinity0> did someone say anything to me?
[12:42] <toad_> sanity_: i'm continuing to learn about db4o's api and db4o's bugs ...
[12:42] <toad_> :)
[12:42] <toad_> requests are 90% working, inserts are 90% working
[12:43] <sanity_> toad_: do you think db4o is going to do what we need?
[12:43] <toad_> freesite inserts aren't tested yet iirc, downloads.dat.gz migration works for requests, not tested for inserts
[12:43] <toad_> yeah, we can definitely do what we want with db4o
[12:43] <toad_> if we can do it with a database, we can do it with db4o
[12:44] <sanity_> toad_: will it address the cpu/memory problems?
[12:44] <toad_> the questions w.r.t. whether a database is a good approach are a tradeoff between more I/O, garbage collection and complexity, versus (potentially much) lower static memory consumption, which in some cases will save us from constant GCing and will generally improve reliability and usability
[12:45] <toad_> and also no startup spike
[12:45] <toad_> and faster startup, especially compared to any other way we could do request resuming
[12:47] <toad_> "no startup spike" results from any form of request resuming
[12:48] <toad_> but loading it from a SimpleFieldSet would cause long startups, high memory usage during the actual startup process, and also during checkpointing, etc
[12:49] <nextgens> infinity0> in fact it's not possible at all: there is still no KeyStore implementation in classpath :<
[12:52] * nextgens is trying to understand the black magic going on in PeerManager.closerPeer()
[12:52] <toad_> :)
[12:52] <nextgens> the obvious thing I could do is to add yetAnotherLoop
[12:52] <nextgens> and to iterate over each of our peer's peers
[12:53] <nextgens> but that would obviously sucks complexity wise
[12:55] <infinity0> nextgens: er what's this then: http://developer.classpath.org/doc/java/security/KeyStore.html
[12:55] * caytchen (n=caytchen@) has joined #freenet
[12:57] * sanity (n=ian@) Quit (Read error: 110 (Connection timed out))
[12:57] <toad_> nextgens: hmmm
[12:58] <toad_> nextgens: why not add getBestLocation(double target) to PeerNode ?
[12:58] <toad_> and then call that?
[12:58] <toad_> that would check the node's own location, and those of its peers
[12:58] <toad_> then you can keep the rest of the logic almost unchanged
[13:02] <toad_> nextgens: w.r.t. backoff, i think we should only exclude (in the location data sharing) nodes that are backed off long-term / have a percentage backoff over some threshold, make sense?
[13:02] <nextgens> infinity0> it throws a notImplemented exception when we use it
[13:03] <nextgens> toad_> I dunno
[13:03] <FreenetLogBot> r21165 (1153) was built successfully on emu, mirrors are updating
[13:03] <nextgens> I'm not the one who simulated the FOAF thingy
[13:03] <toad_> nextgens: did you do getBestLocation(target) ?
[13:04] <toad_> well any simulations are unlikely to have backoff
[13:04] <nextgens> no
[13:04] <toad_> what did you do?
[13:04] <nextgens> I'm comparing the result of Location.distance and trying to minimize the diff
[13:04] <nextgens> arguably it sucks as I compare twice the same thing
[13:04] <infinity0> oh right ok
[13:04] <nextgens> including "my own location"
[13:04] <nextgens> that's a bug I guess
[13:07] <nextgens> can I compare directly two doubles if they are unchanged?
[13:07] <nextgens> hmm
[13:07] <nextgens> not sure about that
[13:07] <nextgens> in fact the check doesn't make sense
[13:08] <nextgens> as it will be pruned by the MaxDiff thingy
[13:08] <nextgens> okay
[13:08] <nextgens> so in fact it should work
[13:08] <nextgens> toad_> could you review what I've commited please? my last three commits
[13:09] <toad_> nextgens: there are lots of twiddles you need to take into account - backoff is just one of them
[13:10] <toad_> do you maintain the existing logic?
[13:10] <nextgens> I think so
[13:10] <nextgens> as it doesn't make sense to propagate the full set of informations
[13:10] <nextgens> right now I publish every routable peers
[13:10] <toad_> hmmm?
[13:10] <toad_> oh
[13:10] <toad_> I mean in the route method
[13:10] <nextgens> and broadcast updates as it changes
[13:11] <nextgens> we can't know about the remote backoff status
[13:11] <nextgens> hence it doesn't make sense to check it there...
[13:11] <toad_> yeah but we have to take the local backoff status into account
[13:11] <toad_> and also per node failure tables
[13:11] <nextgens> the only thing we do is to take the "local" backoff into account: the backoff to the middle man
[13:11] <nextgens> hmm
[13:12] <nextgens> the code is plugged right after the one checking the backoff
[13:12] <toad_> hmmm
[13:12] <nextgens> I think it preserves the logic
[13:12] <toad_> iirc we have a bunch of different choices - best not backed off, best ignoring backoff, best from per node failure tables etc etc
[13:13] <toad_> and then we look at them in priority order
[13:13] <toad_> now, if it only changes how we deal with the location, it should preserve the logic...
[13:13] <toad_> okay, that looks like a nice clean patch
[13:14] <toad_> w.r.t. the first patch (21160) ... we publish all routable peers and ignore backoff status, right?
[13:15] <nextgens> yes
[13:15] <toad_> why is FNPLocChangeNotification deprecated?
[13:15] <toad_> oh doh
[13:16] <nextgens> because we are gonna get rid of it post 1154
[13:16] <nextgens> we don't send it anymore
[13:16] <toad_> wouldn't it be more efficient to keep it and use it when only our loc has changed?
[13:16] <nextgens> as we send FNPLocChangeNotificationNew to all nodes post 1153
[13:16] <nextgens> that doesn't worth the code duplication imho
[13:16] <toad_> hmmm maybe
[13:16] <nextgens> the whole broadcasting thingy sucks, really :)
[13:18] <toad_> it would save some bytes, isn't that good?
[13:18] <nextgens> if that's what we want to achieve we should send differential messages
[13:19] <nextgens> two fields: one containing locations to add, one locations to delete
[13:19] <toad_> well it's just that differential for just our loc is easier than differential in general
[13:19] <nextgens> and maybe a "count" too
[13:19] <nextgens> with those three fields we would handle all cases
[13:19] <nextgens> a negative number would mean "resend everything"
[13:20] <nextgens> and otherwise we could cleanly deal with every cases
[13:20] <nextgens> hmm
[13:20] <toad_> sounds ok
[13:20] <nextgens> no, we would need a "resend request" message too
[13:20] <nextgens> but that's not a big deal
[13:20] <toad_> yay 1.5 code!
[13:20] <toad_> cool
[13:21] <nextgens> in terms of complexity that part of the code really sucks
[13:22] <nextgens> I wonder how many times per second we call PeerManager.closerPeer()
[13:22] <nextgens> but we could greatly improve to lookup time
[13:23] <toad_> in what build do you add the PeerManager method to get all the locations?
[13:23] <nextgens> I didn't add it; it existed for probe requests
[13:24] <toad_> ah
[13:24] <nextgens> I've added PeerNode.getPeersLocation()
[13:24] <toad_> ok
[13:24] <nextgens> PeerManager.PeerLocationsDouble() returns only "our" peers' locations
[13:25] <toad_> well we should consider whether to exclude long-term-backed-off nodes and nodes with % not backed off under some threshold
[13:25] <toad_> apart from that it looks good
[13:25] <nextgens> is that something you want to consult oskar about?
[13:26] <nextgens> or shall I pick some arbitrary value up and implement something?
[13:26] <toad_> it might be a good idea to consult oskar, but i dunno how much of a clue he will have
[13:27] * srivatsan (n=chatzill@) has joined #freenet
[13:29] <toad_> just pick some arbitrary values - exclude it if it's 95% backed off, and/or it is backed off and its backoff period is at least say 10 minutes; we can notify when it comes out of backoff; then the question would be how to ensure that we do notify when the situation improves...?
[13:33] <sanity_> toad_: are you aware of this: https://visualvm.dev.java.net/index.html
[13:33] <sanity_> its a pretty neat monitoring app
[13:38] <sdiz> sanity_: visualvm crashes the jvm quite often when enable cpu profiling
[13:38] <sanity_> i think you need to switch off shared classes
[13:38] <toad_> shared classes?
[13:39] <sdiz> i don't think they are enabled here
[13:41] <sdiz> toad_: shared class allow different jvm instance share the loaded classes..... they just mmap() a file containing the in-memory representation of classes .
[13:44] * toad_ testing micro-darknet on localhost ...
[13:44] <toad_> 2K insert works
[13:44] <toad_> 120K insert works
[13:44] <toad_> 1.37MB in progress
[13:45] <toad_> works
[13:45] * srivatsan (n=chatzill@) Quit (Read error: 104 (Connection reset by peer))
[13:45] <nextgens> toad_> micro-darknet works
[13:45] <nextgens> they are just slow because of the lack of nodes
[13:46] <toad_> why does that cause it to be slow?
[13:46] <toad_> well he said that he had problems with big inserts stalling at 62%
[13:54] * nico_32 (n=user@) has joined #freenet
[13:54] <nextgens> I guess it's because of RNFs
[13:54] <nextgens> each node's location is likely to be really far away from the target location
[13:56] * infinity0 (n=infinity@) Quit (Read error: 104 (Connection reset by peer))
[13:57] <toad_> iirc 2 consecutive RNFs is treated as a DNF, so it will need to RNF 6 times for each block
[13:57] <toad_> that doesn't explain why it stalls for him
[13:57] <toad_> i think there may have been overload, he said something about thread limit rejections
[14:35] <toad_> hmmm, 2 RNFs counts as a success for inserts ... i really don't see why we'd have problems then...
[14:35] <toad_> bbiab
[14:42] * infinity0 (n=infinity@) has joined #freenet
[14:44] * infinity0 (n=infinity@) Quit (Remote closed the connection)
[14:46] * DaffyDuck_ (n=jan@) has joined #freenet
[14:55] <sanity_> sdiz: shared classes are on by default in 1.6 IIRC
[14:55] <sanity_> sdiz: the visualvm documentation tells you to switch them off
[15:00] * Nande (n=KVIrc@) has joined #freenet
[15:17] * Luke771 (n=luke@) Quit ("Leaving")
[15:19] * n0ob (n=travis@) has joined #freenet
[15:23] * sanity_ (n=ian@) Quit ()
[15:29] <toad_> okay, the cooldown queue is definitely working with the current db4o code
[15:35] <nextgens> can testers update to current trunk please?
[15:35] <FreenetLogBot> The monitoring script has detected a network glitch on http://88.198.34.102/~mirror/ : it has been disabled
[15:36] <nextgens> and possibly hack their Version.java to advertize 1154
[15:38] * Dieppe (n=clement@) has joined #freenet
[15:40] * n0ob (n=travis@) Quit ("Now onto something more interesting")
[15:50] <FreenetLogBot> The monitoring script has detected that http://88.198.34.102/~mirror/ is eligible to be back on the main mirror rotation list. Welcome Back!
[16:35] * Nande (n=KVIrc@) Quit ("byes, tengan la bondad de ser felices. ;)")
[16:36] * Nande (n=KVIrc@) has joined #freenet
[16:38] * Mathiasdm (n=Mathias@) has joined #freenet
[17:04] * sanity (n=ian@) has joined #freenet
[17:04] * ChanServ sets mode +o sanity
[17:07] <FreenetLogBot> r21174 (1153) was built successfully on emu, mirrors are updating
[17:08] * TheBishop_ (n=TheBisho@) has joined #freenet
[17:08] * zerwas (i=zerwas@) Quit (Remote closed the connection)
[17:11] * zerwas (i=zerwas@) has joined #freenet
[17:19] <nextgens> doh
[17:19] <nextgens> two mistakes in one commit
[17:20] <toad_> rehi
[17:23] <FreenetLogBot> r21175 (1153) was built successfully on emu, mirrors are updating
[17:30] * Kip (n=KipD@) Quit (Read error: 113 (No route to host))
[17:36] * greycat (i=rfc1413@) has joined #freenet
[17:41] * quadeddie (n=colt@) has joined #freenet
[18:00] <quadeddie> Anybody have a problem with "railties" on OSX?
[18:06] * apterium (n=apterium@) Quit (Remote closed the connection)
[18:06] * Mathiasdm (n=Mathias@) Quit ("Invisible Internet Project: http://www.i2p2.de")
[18:08] * Nande (n=KVIrc@) Quit ("byes, tengan la bondad de ser felices. ;)")
[18:18] * ryanakca (n=ryan@) Quit (Read error: 104 (Connection reset by peer))
[18:23] * quadeddie (n=colt@) Quit ()
[18:23] * ryanakca (n=ryan@) has joined #freenet
[18:26] * Nande (n=KVIrc@) has joined #freenet
[18:27] * archangelpetro (n=hellequi@) Quit ("Build a hundred bridges and suck a single cock, and you're not a bridge builder.. you're a cocksucker")
[18:28] * Johan^mlg (n=bllarf@) has joined #freenet
[18:38] <sanity> ef
[18:40] <reflect> gh?
[18:43] * PrototypeX29A (n=Peter@) has joined #freenet
[18:43] <PrototypeX29A> hi
[18:53] * Johan^mlg41 (n=bllarf@) has joined #freenet
[18:53] * Johan^mlg (n=bllarf@) Quit (Read error: 104 (Connection reset by peer))
[19:01] * skxpl (i=skx@) Quit ("changing servers")
[19:05] * skxpl (n=skx@) has joined #freenet
[19:24] * Kip (n=KipD@) has joined #freenet
[19:47] * DaffyDuck_ (n=jan@) Quit ("Leaving")
[19:49] * saces (n=saces@) Quit (Read error: 104 (Connection reset by peer))
[20:01] * nico_32 (n=user@) Quit (Read error: 113 (No route to host))
[20:02] * saces (n=saces@) has joined #freenet
[20:02] * ChanServ sets mode +o saces
[20:22] * mambro (n=mambro@) has joined #freenet
[20:22] * mambro (n=mambro@) has left #freenet
[20:25] <FreenetLogBot> r21181 (1153) was built successfully on emu, mirrors are updating
[20:25] <FreenetLogBot> r21182 (1153) was built successfully on emu, mirrors are updating
[20:42] * archangelpetro (n=hellequi@) has joined #freenet
[20:43] * Luke771 (n=luke@) has joined #freenet
[20:50] * Luke771 is now known as luke771_afk
[20:51] * luke771_afk (n=luke@) Quit ("Leaving")
[20:54] * zerwas (i=zerwas@) Quit ("Keep on flowing ...")
[21:07] <FreenetLogBot> r21183 (1153) was built successfully on emu, mirrors are updating
[21:27] * zerwas (i=zerwas@) has joined #freenet
[21:38] * greycat (i=rfc1413@) Quit ("This time the bullet cold rocked ya / A yellow ribbon instead of a swastika")
[21:39] <FreenetLogBot> r21192 (1153) was built successfully on emu, mirrors are updating
[21:51] <FreenetLogBot> r21193 (1153) was built successfully on emu, mirrors are updating
[21:51] * nOgAnOo (i=nOgAnOo@) has joined #freenet
[21:54] * Johan^mlg41 (n=bllarf@) Quit ("Miranda IM! Smaller, Faster, Easier. http://miranda-im.org")
[21:58] * Zarggg (n=z@) Quit (Success)
[22:00] <FreenetLogBot> r21194 (1153) was built successfully on emu, mirrors are updating
[22:01] * gregeh (n=gregeh@) has joined #freenet
[22:01] * PrototypeX29A (n=Peter@) has left #freenet
[22:07] <FreenetLogBot> r21195 (1153) was built successfully on emu, mirrors are updating
[22:14] <nextgens> 98% of routing accuracy on a perfect network with FOAF
[22:14] <nextgens> let's see what it gives without
[22:14] <toad_> record the avg path length too
[22:14] <toad_> what was it?
[22:16] <nextgens> I didn't find the metric
[22:16] <nextgens> but a guessestimate would be around 2-3 for 500 nodes
[22:16] <nextgens> it gives the maxHTL
[22:17] <nextgens> but doesn't account hopsTaken
[22:17] <nextgens> that's weird
[22:17] <toad_> hmmm odd
[22:18] <toad_> any ideas for reducing disk access on the db4o branch?
[22:18] <toad_> it's not a lot bytes wise, but less seeks would be nice
[22:22] <nextgens> same accuracy without FOAF
[22:22] <FreenetLogBot> r21198 (1153) was built successfully on emu, mirrors are updating
[22:25] <FreenetLogBot> r21200 (1153) was built successfully on emu, mirrors are updating
[22:26] * zerwas (i=zerwas@) Quit (Remote closed the connection)
[22:26] <toad_> nextgens: how big is the test network?
[22:27] <nextgens> 500 nodes
[22:27] <toad_> hmm that should be big enough for some real routing to happen
[22:27] <nextgens> sure
[22:27] <nextgens> but we have force-disabled swapping
[22:28] <nextgens> and we build an ideal kleinberg network out of it
[22:28] <toad_> ah, on a perfect network ...
[22:28] <toad_> hmm
[22:28] <toad_> there are other tests where we randomize the locations and wait for swapping to stabilise
[22:28] <toad_> that takes a lot longer though
[22:29] * Cooo is going to bed. Godnight nextgens and toad_
[22:29] <toad_> goodnight and thanks Cooo
[22:29] <toad_> 281.463: [GC 44885K->23017K(79232K), 0.0088750 secs]
[22:29] <nextgens> there seems to be a problem with the handshaking code
[22:29] <Cooo> Thanx to you too.
[22:29] <nextgens> it takes way too long
[22:29] <toad_> this is with 2 640M ISOs queued and a load of other stuff
[22:30] <toad_> but i think it will slowly rise until it's constantly GCing... the baseline rises somehow
[22:30] <toad_> it did for Cooo and it did for me before ... but maybe I've fixed it
[22:30] <toad_> nextgens: CPU usage?
[22:32] <nextgens> it shouldn't matter much
[22:32] <nextgens> but of course it's maxed out
[22:34] <nextgens> Average path length for successful requests: 5
[22:34] <nextgens> that's without foaf on a 500 nodes (perfect) network
[22:34] * Zarggg (n=z@) has joined #freenet
[22:34] <nextgens> let's run the same with it
[22:34] <toad_> okay, that's plausible (for a perfect network)
[22:37] <toad_> but we will get the biggest gains on imperfect networks
[22:37] <toad_> and swapping produces imperfect networks
[22:37] <nextgens> I'll run those tests as well
[22:37] <nextgens> but they take way longer :/
[22:38] <nextgens> Average path length for successful requests: 3
[22:38] <toad_> woah
[22:38] <nextgens> okay, that's a definite gain
[22:39] <toad_> it should show it as a floating point number
[22:39] <toad_> that's the next patch :)
[22:39] <toad_> but on the strength of that, feel free to release as soon as you're confident the build isn't completely kaput ... but i do think the longer tests would be interesting
[22:39] <toad_> although they are probably with smaller networks
[22:40] <nextgens> well I'm not confident about my code
[22:40] <nextgens> I've just fixed 2 big bugs since the original commit
[22:40] * gregeh (n=gregeh@) has left #freenet
[22:40] <nextgens> it definitly needs testing
[22:41] <toad_> what were the bugs? the code looked okay
[22:43] <nextgens> I was sending the broadcast way too often
[22:44] * nextgens tries the ulpr tests too
[22:57] * zerwas (i=zerwas@) has joined #freenet
[23:04] <nextgens> the tests have been started
[23:04] <nextgens> will see the results tomorrow
[23:04] * nextgens is off to bed
[23:15] * ahuxley (n=ahuxley@) has joined #freenet
[23:34] * geeku (n=geeku@) Quit ("reboot")
[23:36] * tigerle (n=ella@) Quit (Remote closed the connection)
[23:39] * geeku (n=geeku@) has joined #freenet
[23:42] * albanetcsr (n=Miranda@) has joined #freenet
[23:45] * zerwas (i=zerwas@) Quit ("Keep on flowing ...")
[23:54] * FrinkC (n=FrinkC@) Quit ("Bye")
Irc logs of #freenet : 2008 2007 2006 2005
These logs were automatically created by FreenetLogBot on chat.freenode.net using the Java IRC LogBot.