[05:28] <infinity> IRCsloth: Still around?
[05:29] <infinity> IRCsloth: Can you try a dapper liveCD on that machine to see if it creates the device correctly?
[05:39] <IRCsloth> yeah, I'm here, just not looking at the screen
[05:40] <IRCsloth> I'll try a dapper live cd to see. Fedora core4 wasn't creating the devices for me either so I have a feeling it's a generic udev problem
[05:40] <IRCsloth> RedHat 9 however works like a charm, heh.
[06:05] <infinity> IRCsloth: I just talked to our udev guy who showed me the code in question.  We dfinitelty SHOULD be creating tape devices.
[06:10] <neuralis> infinity: did you ever find anything out about the certification exams?
[06:18] <infinity> Nothing meaninful or interesting, no.
[06:18] <nictuku> considering your experience, what would be an acceptable maximum usage for a network wide updates server, written in python, in mb?
[06:18] <nictuku> nwu, in this case
[06:19] <nictuku> maximum memory usage, I mean
[06:19] <nictuku> like, 10M, 15, 30, 100?
[06:19] <neuralis> what memory-intensive operation(s) are you performing?
[06:20] <neuralis> are you caching something? package lists?
[06:20] <nictuku> neuralis, getting relatively big list of packages and storing in a database. only temporary sqlobjects cache (should end after the thread is closed)
[06:20] <neuralis> so how would you eat up a hundred megs of memory?
[06:21] <nictuku> I mean, not in a developers POV, but in a sysadmin pov.
[06:21] <neuralis> i understand what you're asking, but i'm not sure why you're asking it. why do you think something like nwu will end up chewing up a lot of ram?
[06:21] <nictuku> what would a an admin expect for that? I'm an admin and I don't like how much memory it's using currently, like 24mb
[06:21] <nictuku> neuralis, because it's currently buggy
[06:22] <neuralis> 24mb isn't problematic at all, but it would be nice to cut it down eventually if there's no need for it.
[06:22] <nictuku> I'm fixing that, but although it won't ever use less than 10mb (python VM and all classes loaded), I wonder what would a sane target value for that.
[06:23] <neuralis> remove obvious leaks and memory holes, and then simply don't worry about it.
[06:23] <nictuku> neuralis, last week the memory usage was growing with no limit, until linux oom killer kills the process hehe
[06:23] <neuralis> yeah, leaks are bad, m'kay.
[06:23] <nictuku> ok, I think I'm getting paranoic about that
[06:24] <neuralis> no need.
[06:24] <infinity> As neuralis said, just make sure it doesn't leak.
[06:24] <infinity> Maybe try to make certain bits more efficient, if they're obviously not.
[06:24] <infinity> No one expects a python application to be "slim", however.
[06:26] <ajmitch> I'd never call apt-proxy lightweight, for example
[06:27] <nictuku> that could be a good base of comparision, btw
[06:28] <nictuku> it's using 50mb of memory in a server at work, and it's just sitting there.
[06:28] <nictuku> nwu is fine then :-)
[06:28] <neuralis> nictuku: hoare had something to say about premature optimization... kill leaks and focus on more worthwhile things like security and stability; memory use can be optimized later.
[06:32] <nictuku> it took me a month to spot and stop the leak. i'm hoping that won't happen often after this painful learning
[06:32] <nictuku> but it was fun :-)
[06:32] <ajmitch> hi fabbione
[06:32] <fabbione> morning
[06:46] <nictuku> nwu big inserts: mysql=0m12.316s, sqlite=0m49.179s. some of these in a row (although really unlikely to happen) would kill a small server IO sub-system
[06:46] <neuralis> er. what are these inserts doing?
[06:47] <nictuku> iostat shows a big write queue and usage
[06:48] <nictuku> neuralis, initial machine setup. basically packages versions and repository data
[06:49] <nictuku> oh. transactions off
[06:49] <neuralis> ok, well, it's more than likely that you're using a suboptimal schema.
[06:50] <nictuku> I wouldn't expect a good result now that I remember it's not using transactions.
[10:36] <\sh> moins
[10:36] <\sh> fabbione: actual ubunt--server amd64 daily is broken, right?
[10:37] <fabbione> \sh: dunno.. i am not on server for this release
[10:39] <\sh> well...looks like it...no di kernel modules nothing...trying flight 6 now...debian amd64 port is far from being usable...so I need a running amd64 debian distro...
[10:40] <fabbione> \sh: check the new daily. there was a kernel abi bump
[10:43] <\sh> fabbione: well...using flight6 with updates...should work
[10:43] <fabbione> yes it should
[10:44] <infinity> \sh: The seeds were out of sync, so kernels on all dailies are messed up.
[10:44] <infinity> \sh: We just fixed that, so the next round of dailies will be fine.
[10:46] <\sh> infinity: ok :) but anyhow I need to deliver some ubuntu servers for amd64 just now...so working around is better right now then to wait for the next cd build run :)
[10:46] <infinity> Yeah, just take the daily that built at the same time as Ubuntu Flight 7.
[10:46] <infinity> That one's fine.
[10:46] <infinity> Or start at Flight-6 and upgrade, if you prefer.  Whatever.
[10:47] <\sh> hmm..nice issue right now...trying to install it via ILO and after the graphic boot selection...black screen
[10:51] <\sh> infinity: just for your info: setting vga mode to 640x480x16 and then booting works...the standard setting does't work when you try to install via ILO remote console
[10:52] <\sh> will file a bug after the installation :)
[10:53] <infinity> I'd need access to an ILO-enabled machine to debug it, unless you have more info for me than "doesn't work".
[10:55] <\sh> infinity: no display but running the installer...looks like the switch from graphics mode to text mode works only in 640x480x16 mode...
[11:00] <infinity> Yeah, that's still not enough info to be useful without a machine to look at.
[11:01] <infinity> I don't suppose you know anyone who could give me access to an ILOable box for a bit?
[11:22] <\sh> infinity: sorry no...all the ILOs here are in mgmt networks which are not accessible from the outside:(
[11:26] <infinity> \sh: If you can get me access to just one at any point, I can diagnose and fix the bug, but otherwise, there's probably little point in even filing it.
[11:54] <\sh> infinity: let me try something...
[03:46] <dyn> hi
[03:47] <dyn> problem: at install time I had no network available. later on I edited /etc/network/interfaces but still '/etc/init.d/networking restart' does not pull up eth0. how can I fix this? (the box was installed in server mode so no X11/gui clients are available)
[03:47] <dyn> how can I fix that to work correctly?
[08:22] <lionelp> idimmu ?
[08:25] <idimmu> hi
[08:25] <lionelp> hi
[08:26] <lionelp> I just fill the page on the wiki: https://wiki.ubuntu.com/BinaryDriverHowto/EiconDiva
[08:26] <lionelp> You can have a look, and comment / make changes :)