| systemdlete | If I receive a dpaste'd file but I don't get a url, but instead I get "dpaste:AA13B154" what do I do with it? I tried passing it to the -g option of dpaste, but nothing | 00:24 |
|---|---|---|
| systemdlete | I've read the man page, and even did some ddg on dpaste. It was much easier to use when it spit out a simple URL! | 00:24 |
| systemdlete | (And yes I learned it uses DHT, but what I want to know is how to use the cmd line tools) | 00:25 |
| systemdlete | apologies for crabbing about this, but the docs are kinda slim. | 00:26 |
| * systemdlete longs for the days of simple unix command line tools, with some documentation... | 00:26 | |
| systemdlete | so... what is the secret sauce here? | 00:27 |
| systemdlete | anyway, I continue to TRY to get apt working again on some of my systems here, and this is what I see when I try to track down just one of the fetches: https://paste.debian.net/hidden/fef0977b/ | 00:32 |
| systemdlete | that last bit just hangs, seemingly forever | 00:33 |
| rrq | by my guess, the dpaste man page suggests it needs --aes-encrypt | 01:35 |
| rrq | but the whole thing looks like an over-complification to me as well | 01:36 |
| systemdlete | so, any ideas on why wget fails? | 01:37 |
| systemdlete | rrq, I have checked the time on the system against time.gov | 01:37 |
| systemdlete | on a different system, I tried the same IP address (by substituting the IP address actually used by the wget on the system that failed): https://paste.debian.net/1333762/ | 01:48 |
| systemdlete | so, here, I get a 404. ??? | 01:49 |
| systemdlete | now, what am I overlooking here? something stupid, I am sure. Usually is. | 01:49 |
| rrq | if the server has virtual services then it needs to be accessed with proper domain name; not just IP | 01:59 |
| rrq | apparently domain name ftp.fau.de (has address 131.188.12.211) is the one registered for mirroring | 02:03 |
| rrq | and that responds with 404 still | 02:04 |
| onefang | It's failing apt-panopticon as well. | 02:09 |
| onefang | No 404 though, just missing a package. | 02:10 |
| onefang | Other things are found. | 02:11 |
| onefang | Only on IPv4 though, works fine on IPv6. 2001:638:a000:1021:21::1 | 02:12 |
| * onefang pulls it off the DNS-RR for now. | 02:13 | |
| onefang | As for secret sauce ... I mix Vegemite (tm) with kangaroo mince and ... oh, not what you where asking about. | 02:15 |
| systemdlete | sounds tasty though | 02:16 |
| rrq | .. and now it's working again... | 02:17 |
| systemdlete | oh? let me try it... | 02:17 |
| systemdlete | I'm still getting 404 | 02:17 |
| systemdlete | (same IP) | 02:17 |
| onefang | Maybe their IPv4 is just being flakey? | 02:17 |
| systemdlete | I dunno. | 02:18 |
| onefang | You get 404 on the base page, not just random files? | 02:18 |
| systemdlete | InRelease file | 02:18 |
| onefang | And can you try IPv6? | 02:18 |
| systemdlete | I have ipv6 disabled around these parts... | 02:18 |
| rrq | you need to use the domain name deb.devuan.org and make sure it resolves to th IP | 02:19 |
| onefang | I need to get it enabled at home so I can test apt-panopticon IPv6 testing at home. | 02:19 |
| systemdlete | the domain name will be resolved to an IP before the packet with the request ever leaves my local network. | 02:19 |
| systemdlete | and the IP could change, I know | 02:20 |
| rrq | yes but the http request will carry the domain name in hte Host: attribute | 02:20 |
| onefang | Except apt will include the domain name IN the request to the server, so the server knows which virtual domain to use. | 02:20 |
| systemdlete | the IP address must be a part of some domain, so it could be detected | 02:21 |
| onefang | On my server I have one IPv4, and many domains. | 02:21 |
| rrq | the IP is one of the results of resolving deb.devuan.org | 02:21 |
| systemdlete | hmmm. | 02:21 |
| systemdlete | well, ok. so my test is not very thorough then | 02:22 |
| onefang | So THAT might be your secret sauce, you are not telling 131.188.12.211 which virtual server you want. | 02:22 |
| systemdlete | the problem is, how can I figure out which of the many servers I am hitting | 02:22 |
| rrq | the http request has at least 3 lines: GET /merged/dists/daedalus/InRelease HTTP/1.1^M^JHost: deb.devuan.org^M^J^M^J | 02:23 |
| systemdlete | well, that was only my test case. The secret sauce is a different matter. | 02:23 |
| rrq | send that with: nc 131.188.12.211 80 | 02:23 |
| systemdlete | It sounds to me that the issue is more ipv6 vs ipv4 or something | 02:23 |
| systemdlete | I am able to nping it | 02:24 |
| onefang | The fact is that 131.188.12.211 is currently having issues, so maybe test with something else? Or wait until the script runs that updates the DNS-RR to remove that one. | 02:25 |
| systemdlete | ok, give me a different IP | 02:26 |
| systemdlete | btw, running nc with that address gives me a hang | 02:26 |
| systemdlete | just fyi | 02:26 |
| rrq | did you pass in the 3 lines? .. it may hang afterwards because the server can accept further http request | 02:27 |
| rrq | 02:28 | |
| systemdlete | I am using netcat-traditional vs the openbsd version | 02:28 |
| systemdlete | oh, ok | 02:28 |
| systemdlete | hold on, let me try that | 02:28 |
| * systemdlete forgot that nc waits on stdin | 02:28 | |
| rrq | not thet ^M and ^J should be the control characters | 02:28 |
| rrq | note | 02:28 |
| systemdlete | bad request; "your browser sent a request that this server could not understand." | 02:30 |
| systemdlete | Yes, I converted the nl-cr to newlines | 02:30 |
| rrq | may need actual DOS line endings | 02:31 |
| rrq | (^M is carriage return and ^J is linefeed) | 02:31 |
| systemdlete | ok, that worked | 02:32 |
| rwp | Just passing by, like the low level action as that's best for understanding, but maybe wget would be easier? | 02:36 |
| golinux | rrq has magic powers | 02:36 |
| rwp | Try: wget -O- -q -S --header="Host: deb.devuan.org" http://131.188.12.211/ | 02:36 |
| systemdlete | but note that is on a system that has been getting the problem intermittently. I will try it on a system that gets consistent failures. | 02:37 |
| systemdlete | rwp: I started out doing just that, in fact. | 02:37 |
| rwp | That wget --header option is a way to force a server name with any IP address. | 02:37 |
| systemdlete | but your --header thingy might just do the trick, thanks | 02:38 |
| rwp | For SMTP I really *LOVE* the swaks command. It's awesome! FTW! I wish there was a similar utility for use with HTTP. wget/curl are both useful but not really in the same league as swaks when it comes to the debugging using them. | 02:38 |
| systemdlete | lol. Love the enthusiasm, rwp | 02:39 |
| systemdlete | I got that feeling when I discovered restic and fossil, exactly! | 02:40 |
| rwp | I am going to apply some of that enthusiasm to dinner. Enjoy! :-) | 02:40 |
| systemdlete | bon appetit! | 02:40 |
| systemdlete | (and be sure to have some of that "secret sauce" ^^^ ) | 02:40 |
| systemdlete | I wonder if that one line passed to --header is sufficient. I tried passing those 3 lines rrq gave me (in a file) to -i option of wget instead. But I am not sure that is 100% equivalent | 02:49 |
| systemdlete | when I tried the --header on the target system (the one having consistent problems with apt update) | 02:49 |
| systemdlete | it hung | 02:49 |
| systemdlete | but the -i option does not work any better | 02:49 |
| systemdlete | interesting, rwp's wget options make it work ok here on the system where I don't see persistent problems. But doesn't work on the target system... | 02:51 |
| systemdlete | I can browse the web from the target system without difficulties. So I think the network here is ok, sans ipv6, of course. | 02:52 |
| systemdlete | maybe we can approach this a different way: Why am I seeing timeouts? Is there a way to override the timeouts from the client side? I tried doing that in the apt.conf, but that didn't help. I think it might be an issue on the server side. | 02:54 |
| systemdlete | what is odd is that the number of hops from the target system is actually LESS than elsewhere here | 02:55 |
| systemdlete | so if the problem were latency here, I'd think I'd be seeing the opposite problems | 02:55 |
| systemdlete | rrq, onefang: Is there any way to see the timeouts on the server side? Maybe figure out what conditions are causing this? | 03:00 |
| onefang | Maybe just skip all of this and pick a different mirror? | 03:03 |
| systemdlete | I thought that was what the round robin did? | 03:04 |
| systemdlete | I'm talking about when I use "apt update" not specific servers | 03:04 |
| onefang | The "round robin" tells apt it has a variety of IPs to choose from. How it chooses is up to what ever apt client you are using. | 03:05 |
| onefang | But you have been concentrating on this one mirror. | 03:06 |
| systemdlete | ^^^ | 03:06 |
| systemdlete | I'm back to just working with apt the normal way. It still fails | 03:06 |
| onefang | And that one mirror is currently not part of the DNS-RR, coz I removed it earlier this morning. | 03:06 |
| systemdlete | IOW, instead of going down this garden path, I'm back to just trying to run "apt update" | 03:07 |
| systemdlete | onefang: Can you see individual requests on the servers? Can you tell why connections are failing? | 03:08 |
| onefang | I can only do that on servers I run, which is sledjhamr.org, and a couple others I have accounts on. Coz that involves looking at web server log files. | 03:09 |
| systemdlete | so, what if we set up a test where you give me an IP and the approprirate headers for a wget --header or wget -i | 03:10 |
| systemdlete | unless you are too occupied to do that atm | 03:10 |
| systemdlete | I don't mean to pull you away from other work | 03:10 |
| onefang | I'm still trying to catch up on lots of stuff that the constant house moving and shit stopped me doing for the last couple of years. | 03:11 |
| systemdlete | onefang, I am tooooo familiar with that issue. For a few years, I was moving every few months | 03:12 |
| onefang | apt-panopticon uses curl instead of wget, and it does all this sort of deep dive into things. | 03:12 |
| systemdlete | does that impact how /I/ am calling to the repos? | 03:14 |
| systemdlete | sorry for my lack of understanding of this deep dive | 03:14 |
| onefang | apt-panopticon is my script for testing package mirrors. http://veritas.devuan.org/apt-panopticon/results/Report-web.html | 03:14 |
| systemdlete | oh, I see | 03:15 |
| onefang | So it's full of examples of the sorts of things you are trying to do here. But using curl instead of wget. | 03:15 |
| systemdlete | well, I can use curl instead. I have both | 03:15 |
| onefang | And I wasn't planning on working on it today. Source code is listed at the end of that page. Sorry about the lack of docs and stuff. | 03:17 |
| onefang | Just discovered GIMP has a MIDI control. I can create images while I learn piano keyboard? Will have to try this. | 05:54 |
| n4dir | onefang: make sure to report how that worked | 05:54 |
| onefang | Ah, you can connect any of the basic MIDI note and controller signals to trigger any menu, and many of the tool parameters. Though there isn't any setup by default. | 05:58 |
| n4dir | i meant: if it was fun and all. - But as answered: You first have to asign the midi controllers? You can't just play the piano and tell gimp "heck go for it" ? | 05:58 |
| onefang | That's correct, first have to tell GIMP what a middle C does, then the next note, then .... | 05:59 |
| n4dir | still very nice and all, but not the right stuff for me. Very cool. Good info | 05:59 |
| n4dir | i don't know how to use gimp at all, so to me, too difficult. Report back, if you get anywhere. Very interesting | 06:00 |
| onefang | No doubt there's people out there that have concocted interesting configurations for this. | 06:00 |
| onefang | I'm mostly poking at GIMP now coz they changed how everything works. Which tripped me up when I needed to quickly crop a photo for my girlfriend. | 06:01 |
| n4dir | onefang: what i was searching for is a really good visualizer for "sound" in general (be it midi or audio-recordings a-la ogg). Didn't find one. | 06:02 |
| n4dir | something psychedelic | 06:02 |
| n4dir | but also something you just load and it goes, it is not important enough for me to figure out, configure, and what not | 06:03 |
| onefang | I know there's plenty of that sort of thing. I have installed a bunch, but not used them yet. | 06:03 |
| n4dir | there is plenty, but most i've seen are really basic | 06:03 |
| n4dir | "projectm-jack" is standalone, so i was hoping for it to be nice, but a) i could't configure it and b) the output was rather basic too | 06:05 |
| onefang | Soooo, previous versions of GIMP I just click on a tool, and it lets me apply it to the image. Now, nothing. Is there some sort of new mode thing that's well hidden that let's me actually use GIMP? | 06:06 |
| onefang | On my other computer I loaded up all the screensavers and just let it run. NOthing moves to music, but some are fairly psychedelic. | 06:06 |
| n4dir | yeah, something like that i look for. While we talked i remembered about screensavers. Thanks, will give it a try | 06:07 |
| n4dir | good luck with your gimp problem | 06:07 |
| onefang | Ah, graphics tablet works, just not the mouse. | 06:11 |
| onefang | Ah, "Core Pointer" was enabled, but I had to separately enable "A4Tech USB Optical Mouse", and now I can draw and stuff with my mouse. | 06:14 |
| onefang | There's other interesting input devices. I can use my joystick, and something called "Generic USB Audio Consumer Control". I can sing pictures? | 06:15 |
| onefang | Next up, getting logcheck to behave. | 06:23 |
| systemdlete | I may be on the verge of cracking the secret: If I call "apt -o Acquire::ForceIPv4=true" it works on the systems where it was failing. | 06:53 |
| systemdlete | I can set the defaults in the appropriate files under /etc/apt, I know. | 06:54 |
| systemdlete | But that isn't the point. apt update and friends work on the other systems, and I have not modified their apt configs at all. | 06:54 |
| systemdlete | So, apparently, this has *something* to do with apt/apt-get's choice of IP versions. | 06:55 |
| systemdlete | the systems I am comparing are both daedalus | 06:55 |
| systemdlete | (but a chimaera system also has problems) | 06:55 |
| systemdlete | And, now, I am wondering if my apt-cacher-ng problems might be related to this as well. | 06:56 |
| systemdlete | Is it possible that apt is choosing ipv6 first and failing to try ipv4 upon failure? | 06:56 |
| onefang | Works fine on my IPv4 only home system. | 06:57 |
| systemdlete | and why does it vary from system to system even when the /etc/apt areas are identical (I've checked this!) | 06:57 |
| systemdlete | Most things work just fine here, most of the time, at least on the majority of systems. | 06:59 |
| systemdlete | Besides the files in /etc/apt, are there other files under, say, /usr/share, that might be setting defaults for apt? | 06:59 |
| systemdlete | (some utilities in Linux work that way) | 06:59 |
| systemdlete | onefang: have you added or modified any options under /etc/apt? (just asking) | 07:02 |
| systemdlete | in particular, have you added/modified Acquire::ForceIPv4 to be set to true? | 07:03 |
| onefang | Just searched for that, didn't find it, so no I haven't set that option. | 07:07 |
| onefang | Looks like logcheck is behaving now. The timestamp for rsyslog changed format is what the problem was, so didn't watch my old custom rules. Time for a break before I fix up the next annoyance. | 07:10 |
| onefang | s/watch/match/ | 07:10 |
| systemdlete | so I set the option in a new file in /etc/apt/conf.apt.d to default to ipv4, as per instructions at https://www.cyberciti.biz/faq/howto-use-apt-get-with-ipv6-or-ipv4-transport-on-ubuntu-debian/ but it doesn't seem to be picking up that file for some reason | 07:11 |
| systemdlete | stranger and stranger. The -o option for ipv4 works for upgrade but not for update | 07:17 |
| systemdlete | almost like it is... idk, ignoring it altogether? | 07:17 |
| onefang | The GIMP MIDI controls basically map to the same stuff that you can map the typing keyboard keys to. Which is most things except actually drawing. So I could set things up that playing a particular chord selects the pencil tool, with a specific colour and line thickness, only the level of blue is controlled by the second knob on the left of my MIDI keyboard. A different chord selects something different | 08:06 |
| onefang | . | 08:06 |
| n4dir | way cool | 08:07 |
| onefang | So now I need way more practice. Not only do I have to sing and play keyboards at the same time, but now I gotta draw pretty pictures as well. lol | 08:09 |
| n4dir | i kinda got used to sing while playing the piano, though it sure distracts me from the piano, and i sure don't sing well | 08:11 |
| n4dir | to me it is kind of a practice to get more comfortable with the piano, to play under stress | 08:11 |
| n4dir | i won't add drawing pictures though | 08:12 |
| onefang | I used to play flute long time ago. So now I'm not used to playing and singing at the same time. But this is off topic now. | 08:12 |
| n4dir | yup, sorry to anyone who bothers this | 08:12 |
| medoc | Hello, would there be someone around familiar with devuan use of /sys/fs/cgroup ? | 19:16 |
| gnarface | medoc: stick around, slow channel but someone probably knows about it, yea | 19:20 |
| medoc | Thanks for the encouragement :) | 19:21 |
| FatPhil | medoc: it's no different from any other distro's cgroup. So badly designed to be underfunctional, and then badly redesigned to have overfeaturitis. | 19:36 |
| FatPhil | I did once use it, in order to stop my browsers from taking too much CPU and RAM. Didn't help the responsiveness of my machine. Which was what I trying to use it for. Not looked at it again since, as, from first inspection, it's not just a doubly-badly-designed nightmare, it's also useless too. | 19:39 |
| medoc | What I'm wondering is if there is any scheme envisionned for application use or if it's ok to just create stuff in the root. | 19:46 |
| medoc | I don't think it's useless by the way, it's quite useful to control the CPU intensity of an app and there is no real alternative that I'm aware of (thinking of using it for the recoll indexer) | 19:47 |
| medoc | Not sure why you failed with the browsers, it seems to to work quite well at keeping the indexer at xx% cpu | 19:48 |
| Xenguy | Interesting, I've been using 'cpulimit' lately to try to tame one of my applications | 19:50 |
| systemdlete | I also tried cgroups, just like FatPhil, for the exact same reason: My firefox would blow up in size, even when I was not actively doing anything. cgroups did not help, just as FatPhil says | 19:50 |
| systemdlete | Xenguy, how has that been working for you? | 19:51 |
| systemdlete | I might try it if it actually works. | 19:51 |
| medoc | Had a look at cpulimit, but it's really hacky, using SIGSTOPs to suspend the app, I'd prefer to let the kernel do it properly | 19:51 |
| systemdlete | @medoc: +1 on that | 19:51 |
| Xenguy | systemdlete, My testing was not exhaustive, but I *seem* to have better results if I apply cpulimit to a specific PID | 19:52 |
| Xenguy | systemdlete, e.g.: cpulimit -p ##### -l 25 | 19:53 |
| Xenguy | I also run it as root | 19:53 |
| medoc | For the people who did try the cgroups, did you envision some scheme, e.g. a per user subtree or whatever to make things look a bit orderly ? Or just random directories under the root ? | 19:59 |
| * systemdlete thinks it is time for a reboot of operating systems design, starting with a micro-kernel and relying more on userland via implementation of a *SIMPLE* multi-tiered privilege system | 20:03 | |
| systemdlete | (but this belongs in OT, not here... just sayin') | 20:03 |
| gnarface | ... i still just use "nice"? | 20:09 |
| gnarface | was there anything wrong with that? | 20:09 |
| gnarface | i just gave up on leaving firefox running though | 20:09 |
| systemdlete | yeah, I noticed that ff just keeps gobbling space, never releasing all the additionally-requested memory back to the heap, and then there are numerous threads running, each of which might be eating pieces of memory, so it was hard for me to figure out exactly where the problem was | 20:35 |
| rwp | I think that by default Firefox and Chromium both default to thinking they can use half of your RAM. And Linux kernel will use half of your RAM for file system buffer cache which is a good thing. The result is that Firefox/Chromium effectively have consumed all of the RAM that is available for userland processes. And for many people the web browser is their ONLY app that they have running so that's okay for them. | 22:22 |
| medoc | @gnarface nice does nothing for you if the goal is to prevent your laptop fans to spin up | 22:47 |
| rwp | CPU nice priority means that higher priority tasks will have more cpu time. But what if there are no other process tasks to run? In that case the nice'd task is getting full cpu anyway. | 22:49 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!