Conversation with #inferno at Thu Mar 17 22:22:29 2011 on powerman-asdf@irc.freenode.net (irc) (23:26:29) Fish- left the room (quit: Quit: So Long, and Thanks for All the Fish). (23:26:35) Fish- [~Fish@9fans.fr] entered the room. (23:47:44) k3nt left the room (quit: Ping timeout: 264 seconds). (23:57:37) sl: mjl-: does your mercurial port for inferno work, currently? (23:59:31) mjl-: depends on what you want :) (23:59:35) mjl-: it can read the repositories (23:59:46) mjl-: www.ueber.net/code/ is served by hgfs (00:00:03) mjl-: the write support is not good enough, at least i don't use it :) (00:25:39) sl: heh. (00:26:25) sl: bichued's mercurial port for plan 9 is similar. i can do pulls but it doesn't seem to work when trying to push to bitbucket. (00:29:33) Fish-: sl: can you commit/branch/merge with it? (00:33:34) sl: Fish-: commit, yes; haven't tried branch or merge (00:34:13) sl: when pushing to bitbucket, immediately after entering my username, i get this: (00:34:17) sl: abort: could not import module msvcrt! (00:35:49) Fish-: ok (00:51:48) k3nt [~k3nt@66-169-234-69.dhcp.snlo.ca.charter.com] entered the room. (00:52:47) mjl-: Microsoft Visual C++ Run-Time? (00:53:06) mjl-: plan 9 is actually a variant of windows! (00:54:20) mjl-: i've used the plan 9 port of mercurial to push, but probably only over the filesystem (remote 9p) (01:04:08) vsrinivas: :) (01:09:42) Fish- left the room (quit: Quit: So Long, and Thanks for All the Fish). (02:26:44) anth: i've used the p9 hg to push to bitbucket without issue (02:27:22) anth: which might not actually be helpful, since, not having had any issues, i don't know that i can tell you what to look at to fix yours. (02:50:48) KillerX left the room (quit: Quit: KillerX). (04:31:11) coolesd [4b4af89f@gateway/web/freenode/ip.75.74.248.159] entered the room. (04:31:38) coolesd: two for eight (04:36:28) coolesd left the room (quit: Quit: Page closed). (04:37:47) sl: anth: were you also using bichued's python? (06:00:43) anth: i believe so. that's what i have installed now, but i can't say with 100% confidence that's also what i had installed then. (09:27:07) robot12 [~kazzhilki@proxy10.ts.fujitsu.com] entered the room. (09:46:08) k3nt left the room (quit: Remote host closed the connection). (10:03:32) bvalek2 [c3e41404@gateway/web/freenode/ip.195.228.20.4] entered the room. (10:06:22) Fish left the room (quit: Quit: So Long, and Thanks for All the Fish). (10:06:53) Fish [~Fish@exo3753.pck.nerim.net] entered the room. (11:44:44) mjl-: sl: btw, my limbo mercurial code is not so bad. it could be fixed up so it's safe to do writing. (11:44:58) mjl-: but such things always take more work+time than you hoped :) (12:10:37) mjl-: vsrinivas: btw, did you try getting that libc stuff (realmess.tgz) going on freebsd? (15:23:02) mjl-: hmm, the plan9 programming style is striking again (15:23:12) mjl-: i'm starting to use 1-char data types now... (15:23:21) mjl-: names for data types (15:23:26) mjl-: B: adt ...; C: adt (16:31:47) anth left the room (quit: Quit: anth). (17:01:12) robot12 left the room (quit: Quit: Leaving.). (17:10:28) bvalek2 left the room (quit: Ping timeout: 252 seconds). (19:02:45) vsrinivas: mjl-: nooo :) (20:23:00) mjl-: bummer :) (20:23:11) mjl-: i made a venti/stream, which is venti/copy only it does lots in parallel (20:23:17) mjl-: but on my machine it was slower with a local copy :) (20:23:22) mjl-: slower than venti/copy (20:23:28) mjl-: to a remote location (4ms latency) it is faster (20:23:40) mjl-: but it can't be slower! (20:24:00) mjl-: i also have 100% cpu load, which doesn't seem right (20:26:47) Fish- [~Fish@9fans.fr] entered the room. (20:40:17) mjl-: hah, making some unbuffered channels buffered makes the runtime go from 52 seconds to 36 (20:40:41) mjl-: now lets see what the gc is responsible for. there are lots of allocations & frees here (20:40:56) powerman: mjl-: there is old bug with 100% cpu load while copying large files to remote system (usual mount + usual cp) (20:41:31) powerman: may be related to your case too (20:42:03) mjl-: hmm (20:42:11) mjl-: do you know what the cause for that is? (20:42:17) mjl-: or how i can verify this? (20:42:24) mjl-: verify if it's the same problem (20:42:46) powerman: http://code.google.com/p/inferno-os/issues/detail?id=167 (20:42:49) mjl-: i've got two venti connections. writing requests & reading responses on both (20:43:23) mjl-: powerman-asdf: does the cp make progress? or is it just really slow? (20:44:02) powerman: is there any reasons for such progress? I'm unaware about any fixes related to this… (20:44:44) mjl-: powerman-asdf: but if you see this behaviour, does the cp complete at all? (20:44:59) mjl-: or is it just looping, eating cpu, and never finishing copying the file? (20:45:10) powerman: yep, it's complete AFAIR, just slowly (20:45:34) powerman: no, file is copied fine, but both speed and cpu load are terrible (20:46:32) powerman: you can easily prove is you have same issue using your two servers and setup listed in issue (20:47:06) mjl-: i'll try it (20:48:00) mjl-: hmm, i just tried running the command just on the emu command line. now its down to 24 seconds. so from 36 to 24 just for not having a gui (20:48:03) powerman: if you have same issue with cp, then probably venti hit same bug (and debug and fix it using cp example will be ease than using more complex venti). if not, then it's hard to say anything about venti. (20:48:52) powerman: gui unlikely affect venti… but there may be some background tasks running in gui (like ircfs) which eat cpu..? (20:50:33) mjl-: i had wm/wm, 2 or 3 wm/sh's and a wm/man (20:50:35) mjl-: i suspect the gc (20:50:51) mjl-: perhaps between the read/writes of the network traffic, emu runs the gc (20:51:08) mjl-: and perhaps keeps running a bit while more i/o could be done (20:51:26) mjl-: if you have more memory allocated, because you have more programs, the gc may run even longer (21:28:20) mjl-: powerman-asdf: a cp between the machines does not cause high cpu load... (21:28:22) mjl-: so it's not that (21:28:45) mjl-: with gc disabled it gives the same results. i am up to high system load instead of cpu load (21:28:55) mjl-: still, there seem to be a lot of syscalls going on... (21:29:16) mjl-: i get to 2.5MB/s transfer rate now, which is good enough at first (21:33:54) powerman: mjl-: on what network connection? 1Gbps home network? (21:34:51) mjl-: 50/50 mbit internet :) (21:53:33) powerman: yeah, on 50Mbit 2.5MB/s for inferno looks good enough (22:22:22) The account has disconnected and you are no longer in this chat. You will be automatically rejoined in the chat when the account reconnects.