DRX | I've searched the forum and chat logs, but I don't see anything recent. I'm getting "lzma_read: Read error" on apt-get update. | 04:00 |
---|---|---|
DRX | Specifically, from http://deb.devuan.org/merged daedalus-security/main | 04:00 |
DRX | I've seen this on several systems, each with different DNS. | 04:01 |
rrq | which arch? | 04:03 |
DRX | amd64 | 04:04 |
onefang | That's the same one bb|hcb mentioned couple of hours ago. | 04:10 |
DRX | Maybe on a different channel? | 04:11 |
onefang | Same package, other details where different. | 04:11 |
onefang | Yep. | 04:11 |
rrq | he mention same error; different arch and pkgmaster.d.o specifically | 04:12 |
onefang | Ho was pointing it out to one of the other infrastructure people, who might be able to fix it. | 04:12 |
onefang | Ho -> He | 04:12 |
DRX | Any idea what was wrong (other than a compression / decompression error)? | 04:12 |
rrq | I've tries update on a few hosts and not had the error | 04:12 |
onefang | He did mention same result from other mirrors, but they are copying pkgmaster. | 04:12 |
rrq | tried | 04:13 |
DRX | Maybe it is just some of the mirrors. | 04:13 |
DRX | All the other systems I have seen this in are in the same geographic area. | 04:13 |
onefang | That's what apt-panopticon is for. | 04:13 |
onefang | But I'm trying to relax, so I'm leaving this for the others that are looking at it already. | 04:14 |
DRX | Yeah, probably not much we can do on this one anyhow. I need to relax, too. | 04:14 |
DRX | At least when I update again, I haven't seen the error. Maybe something to do with caching? Probably not a big deal. | 04:16 |
rrq | seems like Packages.xz is gzip compressed | 04:16 |
onefang | Sometimes timing is bad and you catch a mirror while it is updating. | 04:17 |
DRX | That is what I though the first time, but after seeing it 6 times, I think it must be something else. | 04:18 |
Xenguy | onefang, Can't panopticon just tell you if there is a bad mirror? If yes, why wait until tomorrow to check? | 04:51 |
Xenguy | FWIW, 'apt-get update && apt-get upgrade -s' works fine here | 04:51 |
rrq | there is an issue with some Packages.xz files in the repo | 04:55 |
rrq | so some people will run into that "lzma_read: Read error" | 04:56 |
rrq | if updating | 04:57 |
DRX | I was able to upgrade everything that needed to be upgraded, so not a big deal, but it is not an error I had seen before. | 04:57 |
rrq | the Packages file is held in 2 versions: a gzip and an xz compressed.. but right ow there's something wrong with some Packages.xz files | 04:58 |
rrq | end users typically recover automatically by settling for the gz variant, but the xz variant is tried first | 05:00 |
rrq | this seems to be caused by an amprolla issue; we'll know better in a couple of hours | 05:02 |
onefang | Xenguy: apt-panopticon isn't checking everything, every now and then we find a new thing to test for. | 05:17 |
Xenguy | You know best | 05:20 |
onefang | I just checked, apt-panopticon downloads and checks the Packages.xz file, coz it's smaller and I'm trying to keep bandwidth usage down for these constant checks. | 05:22 |
onefang | This problem might explain the errors my apt-panopticons been emailing me for the last day or two while I was- busy moving house. More digging needed, at both ends. | 05:23 |
onefang | Intermittent though, only happens a few times a day. | 05:27 |
onefang | So the sort of thing I do not want to spend my weekend frustratingly tracking down, after my house move from hell, when others have more experience with the actual source of the problem. I'll add a TODO to add atest to apt-panopticon, then try to get back to relaxing So yeah, "wait until tomorrow". | 05:30 |
Xenguy | Put the priority on 'relax' after your travails | 05:31 |
DRX | Yeah, not that big of a deal. Rest up. | 05:32 |
golinux | LeePen made some changes to amprolla that he thought would fix things. | 05:33 |
onefang | People wonder why apt-panopticon does crazy things like check /// in URLs and check the Release files for dates in the future, ... now I'm adding "check if .xz files are really .gz files" ... coz this shit happens. lol | 05:34 |
rrq | I think these xz compressed files have got gzipped (in addition) | 05:35 |
rrq | maybe test for "Packages.* file(s) broken" rather | 05:36 |
onefang | Mostly it's been testing "are the mirrors properly duplicating recent changes on pkgmaster", but I'm happy to add "is pkgmaster itself correct" territory. I already have a TODO for "if pkgmaster fails, don't bother testing the others" | 05:40 |
adam4b | does anyone else have slow connection speed to deb.devuan.org? I have ~200kB/s | 05:57 |
onefang | deb.devuan.org is a DNS round robin, so you'll have to figure out which one you got. | 05:59 |
onefang | Your method of testing the connection speed might show you the IP in use at the time. | 06:01 |
rustyaxe | next release; can i turn my testing install into /stable? My internet situation has become less good and -testing's updates get a little time consuming at 10mbit :P | 06:02 |
onefang | I'm lucky to get double digit KB/s these last few days. lol | 06:04 |
onefang | Welcome to the World Wide Wait. | 06:06 |
adam4b | yeah, | 06:11 |
adam4b | really slow all the time now | 06:12 |
adam4b | rustyaxe that update is exactly what brought me here :p | 06:14 |
onefang | I'm in the middle of moving, and currently using 4G Internet coz I could take it with me. Haven't found the best spot in the new place for it yet. The home has HFC Internet, I just have to get my ISP to turn it on. Want to get the important stuff settled first, fast Internet can come later. | 06:15 |
* onefang can hear some of you thinking "fast Internet is the MOST important!!!". | 06:16 | |
onefang | But I do have other servers testing the package mirrors constantly via apt-panopticon, so if we know which particular server it was... | 06:17 |
adam4b | I don't even mind my 20Mbps, I'm rural, but yes fast internet is gud | 06:17 |
DRX | I still had dial-up less than 15 years ago. Everything seems fast now. | 06:18 |
adam4b | I'm just going by what apt is telling me, idk how to trace the connection | 06:18 |
DRX | netstat, iftop, ... | 06:19 |
onefang | apt or apt-get will just tell you the mirror IP while it's running. | 06:19 |
DRX | Oh, if it is apt, yes it prints it right there. | 06:20 |
adam4b | I'm seeing http://deb.devuan.org/merged | 06:21 |
onefang | apt update prints the IPs at the bottom, but very quickly. lol | 06:21 |
onefang | apt-get did the same thing. I don't have any actual upgrades available to test that. | 06:22 |
* adam4b stops blinking | 06:22 | |
onefang | Or do tho netstat thing. | 06:24 |
DRX | adam4b must have fast internet. | 06:27 |
onefang | Think we already did the speed comparison. lol | 06:28 |
onefang | So if you can find the IP of the DNS round robin package mirror giving you trouble, then https://sledjhamr.org/apt-panopticon/results/Report-web.html might help you dig further. | 06:30 |
onefang | Or you could traceroute the IP and see if there's blockage somewhere between you and it. | 06:30 |
onefang | I'm seeing four servers timing out on that web page now. | 06:32 |
onefang | The other test sites for apt-panopticon are having no timeouts. So central / west Europe maybe? | 06:33 |
DRX | That is a neat page. | 06:36 |
onefang | Maybe the various wars are fucking up the Internet in that general area now that cyperwarfare is popular. | 06:36 |
onefang | Thanks. | 06:36 |
onefang | Of those four timeouts, three are in that general area, one is in Japan. sledjhamr.org is in a centrally located network centre in Amsterdam. | 06:40 |
DRX | Hungary and Berkeley don't look good, either. | 06:41 |
DRX | Shame. | 06:41 |
onefang | Berkeley has looked bad for years. | 06:41 |
onefang | More timeouts! | 06:41 |
onefang | For those playing along at home, https://pkgmaster.devuan.org/mirror_list.txt will tell you where each package mirror is, and the other two apt-panopticon test sites are http://veritas.devuan.org/apt-panopticon/results/Report-web.html and https://ap.in.devuan.org/apt-panopticon/results/Report-web.html | 06:43 |
adam4b | I've never played with netstat before,neat | 06:43 |
onefang | These timeouts are all over the planet this time, but only the test server in Amsterdam is seeing them. | 06:44 |
onefang | Soooo, might be a problem with my test server, which is also my package mirror, which might just happen to be the one you got with the error in the first place, or some Internet problem in Europe. | 06:45 |
DRX | They say it is always DNS, but maybe BGP this time? | 06:46 |
DRX | Or maybe some carriers having a peering tiff. | 06:46 |
onefang | Hmmm, that might also explain my personal network speed issues. I VPN via my server. | 06:47 |
onefang | NOW it's personal! lol | 06:48 |
adam4b | just found panopticon-project.... that's uhh fascinating read | 06:50 |
onefang | Glad you like it. | 06:51 |
onefang | Wait "onefang> NOW it's personal! lol" was NOT the queue for the arrival of the Goat Avenger. | 06:52 |
onefang | Only three timeouts now. | 06:53 |
adam4b | so my night went like, start a DJ set on youtube, decide to try recording in vlc, can't figure out why it buffers, decide to download mpv, do apt update..... now I'm reading P2 | 06:55 |
adam4b | wild | 06:55 |
onefang | I've had a wild ride since November, but that's waaaay off topic. lol | 06:56 |
onefang | The timeouts keep fluctuating, back to six of them. | 06:57 |
adam4b | is there a more useful variation of 'netstat -tv' | 07:04 |
onefang | -vuapt is what I tend to use, sometimes add -n | 07:04 |
adam4b | i'm doing nslookups on the ip but it's tedious | 07:04 |
DRX | I usually add -n, but it should do nslookups without it. | 07:05 |
DRX | I.e. -n suppresses DNS lookups and just gives IP addresses. | 07:07 |
onefang | Also gives port numbers instead of protocol names. | 07:08 |
DRX | With -t, there should only by TCP. It prints it, but it will all be the same with -t unless there are other protocols (like -u for UDP) specified. | 07:10 |
onefang | That's why I usually have the -t and -u options. | 07:11 |
onefang | NO TIMEOUTS! | 07:12 |
adam4b | well I have the connection narrowed down to either devuan.ipacct.in or dev1.ipacct.in | 07:13 |
DRX | Looks much better. | 07:13 |
onefang | Same server, both in India, bb|hcb runs them. | 07:14 |
DRX | Has the one in Hungary been down for a while? | 07:15 |
onefang | quantum-mirror.hu? It's up. | 07:16 |
onefang | It only fails the URL sanity test, which mostly wont be a problem. | 07:16 |
DRX | Ah, that explains the curious name resolution there. | 07:17 |
onefang | "The URL sanity test replaces "/" in URLS with "///", to see if the mirror can cope with that. This might happen due to a minor mis-configuration by the apt user, but decent web servers should cope with that. The result for a mirror that does not cope is a failed download for that user, so this is an ERROR. We care about this coz apt has had multiple bugs in the past where they might let URLs like that slip through." | 07:17 |
onefang | I'm still scratching my head about how you might configure a web server to fail /// in a URL. | 07:18 |
onefang | But four of our mirror admins figured it out. lol | 07:19 |
adam4b | well, good job :thumbs up: | 07:21 |
DRX | 4 haven't figured it out, yes. | 07:22 |
DRX | Just checked the webserver I take care of; it handles "///" just fine. Whew. | 07:23 |
DRX | Good old lighttpd. | 07:23 |
onefang | It used to be more than four, but the others fixed things last time I asked everyone to. | 07:23 |
DRX | Just looking out of interest, it looks like two are Nginx, one is Apache, and one is Caddy (haven't seen that one much). No correlation. | 07:29 |
onefang | Yes, we looked at that before. | 07:30 |
onefang | 80 KB/s, I might make three digits yet! lol | 07:32 |
adam4b | thanks for the tips, gl with the updates | 07:34 |
DRX | Ditto; very handy info on the mirrors, and very nice site. | 07:35 |
onefang | You are welcome. | 07:38 |
raub | When I installed devuan I told it not to link, say, bin -> usr/bin/. If I want to undo that, can I just boot off an iso and do the aliases manually or is there something in addition to that I must do? | 07:45 |
rrq | that should be fine. (copy the files of course) | 08:02 |
guest1 | I'm installing devuan 5 using an existing /home partition, why does rsync process the existing contents of it? There's a lot, so it's taking forever, and I'm not sure it's even necessary | 16:00 |
gnarface | guest1: that's a good question, i'm not sure why it would look there at all... did you do anything weird? and where did you get the devuan image, can you show it? | 16:03 |
guest1 | devuan_daedalus_5.0.0_amd64_desktop-live.iso from files.devuan.org | 16:05 |
gnarface | guest1: ah, maybe it's something to do with the live installer. fyi the other regular installers are more like debian's | 16:05 |
gnarface | actually, that's almost certainly the case, as the live installer just basically installs itself to the harddrive | 16:06 |
gnarface | i'm not sure preserving home directory contents is something it actually does, but i have not used it... | 16:07 |
guest1 | can i force it to skip /target_home ? | 16:07 |
gnarface | that's what i'm saying, i don't know. you can make the other ones not do that, but the live installer has fewer options | 16:07 |
guest1 | I sent sigstop to rsync while checking what happened to the files (nothing, it seems) | 16:07 |
guest1 | it just....takes forever | 16:07 |
gnarface | you probably want this one: https://files.devuan.org/devuan_daedalus/installer-iso/devuan_daedalus_5.0.1_amd64_netinstall.iso | 16:08 |
gnarface | it is much smaller | 16:08 |
gnarface | and if you do manual partitioning you can tell it to skip home | 16:08 |
guest1 | now how do i safely abort yad mid-install | 16:09 |
gnarface | dunno, depends on how far it got... | 16:09 |
gnarface | on the plus side, it also works as a rescue cd :) | 16:10 |
guest1 | if / sorts before _ then i think i should just be able to kill rsync, and i'll remount /target_home in readonly mode just to be safe | 16:11 |
gnarface | if it hasn't touched anything, you should make a backup before you proceed | 16:12 |
guest1 | it made it through my steam folder which shows no modifications '=D | 16:13 |
guest1 | ok unmounting target_home seems to have forced it to skip everything | 16:20 |
gnarface | are you trying to continue the live install while forcing it to skip /home? note that the live iso may give you an install that's not be able to start its graphical environment properly if you do that, i'm not actually sure | 16:21 |
guest1 | yes, i think it might just work, since it shouldn't adjust anything in the home folder anyway | 16:23 |
gnarface | i think that may be a mistaken assumption... | 16:26 |
gnarface | like i said, dunno for sure though | 16:26 |
guest1 | oh, i see, it will try to autologin to devuan, so i have to find that configuration somewhere | 16:26 |
gnarface | well it also starts a graphical login, which on its own may expect certain things to be in your home directory even without the auto-login | 16:26 |
guest1 | it clobbered the apt database, oh well | 16:28 |
gnarface | :( | 16:29 |
gnarface | do you have somewhere to backup the home directory to? | 16:29 |
gnarface | maybe the better plan would be to backup the home directory and let it do what it wants | 16:29 |
guest1 | the home drive is unplugged and safe | 16:29 |
Afdal | hey boyos, how do I flush my DNS cache with sysvinit? | 23:30 |
Afdal | and presumably other inits | 23:30 |
Afdal | I assume system is the only init that's attempted to take over dns cache control | 23:31 |
Afdal | systemd | 23:31 |
Afdal | hmm, is it through dnsmasq? | 23:32 |
Afdal | oh it seems I don't even have that installed | 23:33 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!