systemdlete | There's been a change in postfix from ascii to beowulf. The same configuration no longer works in beowulf. I'd prefer not to have to completely reconfigure what was working fine before. A web poster stated that it was due to (changes in?) cyrus sasl. Istr there was a similar issue in another distro, but I don't recall what I had to change to get it working again. | 00:44 |
---|---|---|
systemdlete | BTW, for some reason, I had to disable ipv6 in postfix on beowulf. That might just be due to my not having a completely deployed ipv6 network here. But even after that, I am getting error when I send mail in the postfix mail log "SASL authentication failed; cannot authenticate to server smtp.gmail.com[173.194.205.109]: invalid parameter supplied" | 00:48 |
systemdlete | Again, this still works fine with postfix on Ascii. | 00:48 |
golinux | Problems like this are why I'm still on jessie | 00:50 |
gnarface | systemdlete: vaguely i recall an issue with the key format change, maybe you just need to regenerate your keys with the new tool version or something | 00:50 |
gnarface | systemdlete: although my google searches suggest this error could also be caused by your ISP blocking port 25 | 00:51 |
gnarface | systemdlete: (blocking port 25 *outbound*) | 00:51 |
systemdlete | I'm talking to port 465 | 00:53 |
gnarface | systemdlete: yikes. another search suggests the same error could be caused by your password being changed at the remote end too. this issue might be generic or it might be gmail specific. | 00:53 |
systemdlete | I can send mail from ascii to the same mail server at gmail. | 00:53 |
systemdlete | Now, the key format change... I'll look at that, thanks. | 00:53 |
gnarface | systemdlete: digging into the search results like grabbing for a raffle ticket i also have this one simple possible fix for you: try adding "smtp_sasl_mechanism_filter = login" to the main.cf | 00:54 |
gnarface | systemdlete: (sorry, never actually used postfix myself) | 00:54 |
systemdlete | np, | 00:54 |
gnarface | systemdlete: (https://bugzilla.redhat.com/show_bug.cgi?id=1575424) | 00:55 |
gnarface | systemdlete: i'm not sure if that's related either, this looks like an unfortunately common type of failure, so the search results are just bombed with it | 00:55 |
systemdlete | I can use plain password from ascii, so I'd think that would still work on beowulf. | 00:55 |
systemdlete | (yes, I noticed that also) | 00:56 |
systemdlete | (amazing how it has NOT been CLEARLY addressed anywhere.) | 00:56 |
systemdlete | (somewhere, an email consultant is making a fortune) | 00:56 |
systemdlete | Brilliant! That worked. thanks gnarface. | 01:00 |
systemdlete | I guess if I had kept on poking through the zillions of search hits, I would have eventually found it. I appreciate your much sharper eyes. | 01:00 |
systemdlete | I did see hits where someone reported that their mail log said "no mechanism found" or something like that. But that is not the same as the error I was having, so I dismissed it. | 01:01 |
* Xenguy loves it when a plan comes together ... | 01:02 | |
* systemdlete hates it when he just didn't observe carefully enough | 01:02 | |
* systemdlete goes back to what he was doing | 01:02 | |
* systemdlete thinks, what was he doing before that... | 01:03 | |
systemdlete | oh, right. | 01:03 |
gnarface | systemdlete: no problem. the key for me was recognizing that the suffix of the error "invalid parameter supplied" was the more relevant clue | 01:09 |
gnarface | systemdlete: (since it appears that cyrus-sasl has a long history of choking on everything) | 01:10 |
gnarface | systemdlete: (i can't even guess who to blame this on) | 01:10 |
systemdlete | cyrus-systemd ? | 01:10 |
gnarface | hehe, no idea | 01:10 |
systemdlete | thanks many times over, gnarface. I am trying harder and harder to figure stuff out on my own and doing more thorough research | 01:11 |
systemdlete | But I still fail it seems. I guess I'll get better. | 01:11 |
gnarface | everyone needs a second pair of eyes from time to time, the human visual cortex uses a lossy compression mechanism | 01:15 |
gnarface | you have some control over it, but it's mostly subconscious so it can be slippery | 01:18 |
gnarface | but you definitely don't want to encourage it :) | 01:18 |
danuan | having random slow speed problems with apt-get update upgrade dist-upgrade | 03:51 |
Xenguy | Use 'deb', not 'pkgmaster' ? | 03:51 |
danuan | can go from 3meg a sec to 80kb 20 kb a sec mid upgrade | 03:51 |
danuan | i can cancel that upgrade go and install just a few pacages and it goes back up to 3 meg a sec | 03:52 |
danuan | yes deb | 03:52 |
Xenguy | Bad deb, bad | 03:52 |
Xenguy | I think I had a bad deb today as well | 03:53 |
danuan | and at same time when it gets slow i can edit apt/sources.lit and switch to tor and it still goes slow | 03:53 |
Xenguy | Packages not found, or something | 03:53 |
danuan | switching to pkgmaster , same thing | 03:55 |
danuan | 80 to 20 kb a sec | 03:55 |
danuan | must be my isp throttling , but why so randomly back and forth , and how in a heck they know to throttle tor at same time when i never used it. | 03:56 |
danuan | and going anywhere else like youtube or speedtests it back as normal 3-5 meg a sec | 03:57 |
Xenguy | Huh, that's weird | 04:03 |
Xenguy | If you use 'mtr', maybe worth a check? | 04:04 |
danuan | mtr ? | 04:04 |
Xenguy | It's a great networking utility that uses traceroute and pings to, er, do a tracetroute : -) | 04:04 |
danuan | traceroute to deb.devuan.org ? | 04:05 |
Xenguy | Or wherever things are slow | 04:06 |
Xenguy | Might give additional info, or not | 04:06 |
danuan | yea last 10 ips do not resolve , just *** | 04:06 |
Xenguy | Huh, interesting | 04:06 |
danuan | but hops before that are all fast sub 100ms | 04:07 |
Xenguy | So some kind of undetermined routing issue perhaps | 04:07 |
danuan | and pings to deb.devuan.org are also sub 100ms | 04:07 |
Xenguy | Sounds like deb is the better option at the moment | 04:08 |
danuan | i am also still on ipv4 , disabled ipv6 on all internal routers as up to now i had no need for it | 04:09 |
danuan | might need to start ipv6 | 04:09 |
Xenguy | I don't speak IPv6 yet either | 04:09 |
danuan | any other sources beside deb and pkgmaster or tor i can try when things slowdown ? besides out of the country , but thats also an option to test | 04:11 |
yanmaani | danuan: the internet has been having problems today, you may want to try again tomorrow | 04:23 |
danuan | this is ongoing for a while , for me atleast | 04:23 |
gnarface | danuan: the international cross-connects seem to be the issue last i checked | 04:28 |
gnarface | danuan: specifically the ones between the US and the UK | 04:28 |
gnarface | danuan: dunno what's up, but it varies wildly | 04:28 |
gnarface | danuan: though, you're also using comcast | 04:29 |
gnarface | danuan: traceroute hops don't have to resolve to work, but they do have to work. try with reverse-lookups disabled and tcp mode enabled | 04:30 |
gnarface | danuan: (then note any hops with abnormally high latency) | 04:31 |
gnarface | danuan: the only official mirrors are the ones in the deb.devuan.org round-robin. we could use a US one badly... | 04:32 |
danuan | wait , if one comp ping resolves deb.devuan.org to mirrors.dotsrc.org and another to megnum.packet-dain.de | 04:41 |
danuan | can that cause weird things midstream if an upgrade and just jump around different hosts ? | 04:42 |
danuan | and .de would imply i am going overseas ? | 04:44 |
gnarface | danuan: first question: no - second question: yes | 04:47 |
gnarface | danuan: well, as to the first question, the mirrors should be all identical in theory. in practice, they may fall out of sync sometimes, but when that happens just retry | 04:47 |
danuan | but drastic speed changes during dist-upgrade start with 3 4 meg a sec and all of a sudden go to 80 - 20 kb a sec | 04:48 |
gnarface | yea, it's brutal. if you have to do multiple machines, use a caching proxy | 04:49 |
gnarface | at first it was bad for me, then it was doing alright for a while, now recently it's gotten bad again | 04:50 |
gnarface | for the most part it's particular to people in the US | 04:50 |
danuan | but why would tor do the same thing , should it not avoid same paths or protocols atleast ? i am puzzled | 04:51 |
gnarface | well, i don't recall tor being known for speed | 04:52 |
gnarface | however that part you can as likely blame on comcast | 04:52 |
gnarface | they have a history of throttling | 04:52 |
gnarface | they have a history of lying about it too | 04:52 |
danuan | but so selectevely , only affecting my devuan upgrades while keeping everything else up to speed , dang | 04:53 |
gnarface | well something like youtube, they can readily access through clusters likely local to the data centers their own servers are in | 04:54 |
gnarface | youtube can afford to put local servers all over the place | 04:54 |
gnarface | and comcast can afford to look good by giving preference to that bandwidth | 04:54 |
gnarface | cheaper for them, faster for you | 04:55 |
danuan | i see | 04:55 |
gnarface | targets the broadest demographic | 04:56 |
danuan | fast porn but slow devuan | 04:56 |
gnarface | only power users notice they're cheapskates | 04:56 |
gnarface | but we don't know for sure whether at this moment the issue is comcast or the international peering | 04:57 |
gnarface | both, unfortunately for you, have a history of bottlenecks | 04:57 |
gnarface | some traceroutes and educated guessing can pretty easily narrow down the actual appropriate entity to focus blame on, but i don't know what else that will profit you | 04:58 |
gnarface | if it's comcast throttling, sometimes it's dumb and you can just change ports | 04:58 |
danuan | am i looking for slow response en route or dropped packets ? | 04:59 |
gnarface | either, both | 04:59 |
danuan | thanx | 05:00 |
gnarface | traceroute -n -T | 05:00 |
gnarface | (try it with and without -T) | 05:01 |
gnarface | (some hops may only respond to one or the other) | 05:01 |
danuan | i will during next upgrades , as to catch the right mirror i am being sent to at the time | 05:01 |
gnarface | also, if the middle hops of the route change a lot when you run it repeatedly, that could be a sign something is wrong out there | 05:02 |
danuan | gnarface , it seems that whenever i am sent to germany things get slow | 05:29 |
danuan | -n without -T for traceroute does not show last 10 hops, * * * | 05:29 |
danuan | with -T a few drops but otherwise fast response | 05:30 |
danuan | would it be wise to hardcode the fast mirrors that i get sent to when doing apt-get to sources.list ? | 05:35 |
danuan | like infinitegrid.org seems closer then germany and fast | 05:36 |
danuan | yea it switches to mirror.checkdomain.de and speed drops | 05:46 |
danuan | midupgrade | 05:46 |
gnarface | danuan: i dunno really but i know that other people have chosen that approach | 08:01 |
gnarface | danuan: you wouldn't be the first | 08:02 |
CAPTCHA_REQUIRED | how do I install devuan on ppc64be or ppc64le? | 11:07 |
r3boot | download the iso for ppc64le and do the install | 11:17 |
crashoverride | hey guys | 15:07 |
crashoverride | is openrc mature in beowulf? | 15:08 |
crashoverride | or is it still only partial? | 15:08 |
fsmithred | it uses sysvinit scripts | 15:08 |
crashoverride | ok | 15:08 |
crashoverride | gracefully? | 15:08 |
fsmithred | check discussions at the forum | 15:08 |
fsmithred | yeah, it works | 15:08 |
crashoverride | ok, hopefully it won't have regressions | 15:09 |
crashoverride | thanks! | 15:09 |
crashoverride | :) | 15:09 |
xrogaan | netbase 6.1 is supposed to be in the backports, but I can't find it through devuan's apt. It's a blocker to the upgrade of iptables. | 17:18 |
xrogaan | Anybody has the same issue? | 17:18 |
fsmithred | yeah, I see that it's not in devuan backports | 17:26 |
fsmithred | but is in buster-backports | 17:26 |
fsmithred | I meant beowulf-backports | 17:26 |
xinomilo | it was added in debian buster-backports in 28/8 | 17:26 |
fsmithred | for devuan | 17:26 |
xrogaan | maybe restate whatever you meant? ^^ | 17:51 |
fsmithred | who restate? We are looking into the problem. | 17:55 |
fsmithred | amprolla needs a kick | 17:55 |
fsmithred | xrogaan, all better: 6.1~bpo10+1 100 | 18:43 |
fsmithred | 100 http://pkgmaster.devuan.org/merged beowulf-backports/main amd64 Package | 18:43 |
fsmithred | might take up to a couple hours to propagate to mirrors | 18:44 |
xrogaan | thanks | 18:44 |
fsmithred | thanks for letting us know | 18:45 |
boardchess123 | hello | 19:42 |
systemdlete | Is there a simple way to find out what changed in one or more packages before upgrading? Does aptitude do this? | 20:58 |
gnarface | other than consulting the changelog i don't know of anything that doesn | 21:01 |
gnarface | that doesn't require manually downloading the package | 21:01 |
systemdlete | I wonder if I can script it, calling apt-file, or some way like that | 21:02 |
gnarface | it would be doable | 21:02 |
systemdlete | I may embark on such an endeavor. | 21:02 |
gnarface | apt-file wouldn't be enough to detect all changes though i don't think | 21:02 |
gnarface | you also have to examine the bundling/unbundling scripts | 21:03 |
gnarface | (some files aren't actually in the archive, but are rather created on the fly by the package's preinst/postinst scripts) | 21:03 |
systemdlete | Thanks again for the postfix help. I actually started reading the millions of options available in postfix. I think one could easily make a living as a postfix SME. | 21:03 |
gnarface | likely | 21:04 |
gnarface | anyway, no problem | 21:04 |
systemdlete | gnarface: Really, all I want to know is what, generally, the intention of the package upgrade was intended. | 21:04 |
gnarface | oh, well the changelog is probably sufficient then | 21:04 |
systemdlete | right. | 21:04 |
gnarface | packages.debian.org should be plumb-able for all the info you need | 21:05 |
user282069 | update-alternatives for gcc won't let me change to gcc- | 21:05 |
gnarface | at least for most the packages | 21:05 |
user282069 | gcc-8; pardon | 21:05 |
gnarface | i dunno about that but there might be some conflicts between gcc8 and later versions that is sabotaging things | 21:05 |
systemdlete | The reason is that I might not want to go through the trouble of, say, rebooting for a new kernel if the change is not urgent. OTOH, if I see a bugfix for something I've been waiting for, I might not want to waste one more minute. | 21:05 |
gnarface | i do recall having to recently purge parts of gcc8 to make room for gcc-9 or gcc-10 on ceres | 21:06 |
gnarface | but yea the changelogs are for that | 21:07 |
gnarface | it's not very useful when they just put "new upstream release" as the reason for the change but when it's a CVE they usually include the number | 21:07 |
user282069 | any idea why with gcc8 installed;;;; i read :::: There is only one alternative in link group gcc (providing /usr/bin/gcc): /usr/bin/gcc-7 | 21:07 |
user282069 | Nothing to configure | 21:07 |
gnarface | user282069: just the vague notion it might be a conflict between two adjacent gcc versions like i was just describing above | 21:08 |
user282069 | thank you very much gnarface // sorry to have been repetitive there. yes it's a neat coincidence | 21:09 |
gnarface | user282069: but gcc is spread out across dozens of packages so maybe you're just missing some or have mismatched versions (like half of one version and half of the other installed) | 21:09 |
user282069 | ah anything is helpful. thank you very much for the clues. | 21:09 |
user282069 | i think i tried to build it intoxicated one night o_0 | 21:09 |
gnarface | dpkg -S /usr/bin/gcc* | 21:10 |
gnarface | what does this tell you? | 21:10 |
user282069 | http://ix.io/2vOr | 21:11 |
gnarface | sorry, if you want me to actually look at the paste use paste.debian.net | 21:11 |
gnarface | i was more asking in general form though | 21:11 |
user282069 | yes of course no problem. thank you | 21:11 |
gnarface | does it only show you gcc-7 or do you see dozens of versions? | 21:12 |
user282069 | it lists gcc 5 6 7 8 and -ar and -nm -ranlib | 21:12 |
gnarface | ok, expected | 21:12 |
gnarface | dpkg -S '/usr/bin/gcc-?' | 21:12 |
gnarface | how about this one? | 21:12 |
gnarface | it should be more specific | 21:12 |
gnarface | is it only showing /usr/bin/gcc-7? | 21:13 |
user282069 | 4 versions where they should be | 21:13 |
gnarface | hmmm, but update-alternatives doesn't see em? | 21:13 |
user282069 | order is 6 7 5 8 | 21:13 |
gnarface | that's weird, i wonder if it's an issue with your alternatives | 21:13 |
gnarface | is this beowulf? | 21:13 |
user282069 | yes | 21:13 |
gnarface | are you fully upgraded? | 21:13 |
user282069 | this afternoon anyway ye | 21:13 |
gnarface | hmmm. | 21:13 |
gnarface | what does your sources.list look like, maybe i should sanity check it | 21:14 |
gnarface | ? | 21:14 |
gnarface | the hostnames should be deb.devuan.org now, people have had problems using the old hostnames and getting outdated responses | 21:14 |
gnarface | beowulf should be pretty stable so this issue suggests an incomplete upgrade or a dependency corruption or something like that | 21:15 |
user282069 | pkgmaster beowulf only;; main non-free contrib | 21:16 |
gnarface | hmm | 21:16 |
gnarface | what about beowulf-security and beowulf-updates? | 21:16 |
user282069 | yes i think i screwed it trying to build myself up a c++17 while on ascii | 21:16 |
user282069 | yes but no backports | 21:16 |
gnarface | hmm, if you have custom-built packages from ascii they could be the culprit | 21:16 |
gnarface | can you uninstall them to try to see if they're blocking legitimate beowulf packages? | 21:17 |
gnarface | your custom package may need a rebuild for beowulf | 21:17 |
user282069 | that's right;; is there any apt or dpkg magic to help me out here | 21:18 |
user282069 | im not sure what it would be;; i had ascii a long time | 21:18 |
gnarface | hmm. well when you try to upgrade with "apt-get update && apt-get dist-upgrade" i would expect it to complain right away | 21:18 |
gnarface | i would expect it to tell you which packages were in conflict | 21:19 |
user282069 | and thank you again for showing me around devuan tonight o/ it's a pleasure | 21:19 |
gnarface | usually it's just a dumb version check issue | 21:19 |
user282069 | yes i have my upgrade notes still i think. it took 3-5 attempts but did finish cleanly | 21:19 |
gnarface | if something went wrong with the pre/post install/uninstall package scripts the damage may be easier to repair with a reinstall, it kinda just depends how far off the rails you went | 21:24 |
gnarface | there's a wide range of possibilities but very little direct evidence so far | 21:24 |
gnarface | if the upgrade really completed correctly it could be as simple as something got automatically removed without being automatically replaced, that you need to manually reinstall the beowulf version of | 21:28 |
gnarface | it's easier to catch that in advance than after the fact, it'll tell you ahead of time what it's going to remove | 21:29 |
gnarface | i do see gcc7 stuff in beowulf still, but it may be picky about it being the right version | 21:33 |
gnarface | specifically 7.4.0 | 21:33 |
user282069 | http://paste.debian.net/hidden/a3cabb21/ my dist-upgrade notes; newest at top i think | 21:35 |
user282069 | nothing stands out :: seemed to finish fine. hrm. maybe i find info on the bad GCC build attempt | 21:37 |
gnarface | user282069: the dkms failure seems kinda bad actaully, i'm surprised it still booted | 21:38 |
gnarface | user282069: did you ever circle back to make it rebuild the initrd.img correctly? | 21:39 |
gnarface | user282069: i don't see how that could be causing the gcc issue, but it also seems like it would have definitely left something broken | 21:40 |
gnarface | user282069: also, packages refusing to disappear after being purged seems like a problem too | 21:41 |
gnarface | user282069: (usually that doesn't happen with orphaned packages unless something very bad happened during the install) | 21:41 |
gnarface | user282069: do you have a live image handy you can boot from for repairs? | 21:42 |
gnarface | user282069: or wait, has this not actually even been rebooted since the upgrade? | 21:48 |
tom_ | Has anyone here successfully installed mariadb 10.5 on beowulf? | 22:28 |
tom_ | i get this error | 22:28 |
tom_ | dpkg: error processing archive /var/cache/apt/archives/mariadb-server-10.5_1%3a10.5.5+maria~buster_amd64.deb (--unpack): | 22:28 |
tom_ | new mariadb-server-10.5 package pre-installation script subprocess returned error exit status 1 | 22:28 |
tom_ | not sure what's going on | 22:28 |
tom_ | oh | 22:29 |
tom_ | nervermind | 22:29 |
onefang | systemdlete: There's already a package that will list the changelogs before asking you if you want to go ahead with the update, then will email them to you. apt-lishchanges. apt-listbugs might be useful for you a well. | 22:32 |
onefang | Warning kernel changelogs tend to be huge. | 22:33 |
systemdlete | apt-lishchanges, or apt-listchanges? :D | 22:33 |
mason | "apt changelog" too | 22:33 |
mason | for a manual approach | 22:33 |
onefang | The second one, I just woke up. lol | 22:33 |
mason | apt-lishchanges is after you've been drinking for a while - it disables root for a few hours proactively | 22:34 |
onefang | lol | 22:34 |
systemdlete | puts your system into a sort-of "promiscuous" state? | 22:35 |
systemdlete | apt-listchanges: APT_HOOK_INFO_FD environment variable is not defined | 22:36 |
systemdlete | (is Dpkg::Tools::Options::/usr/bin/apt-listchanges::InfoFD set to 20?) | 22:36 |
xinomilo | also, `apt changelog package` | 22:36 |
mason | That's not supposed to happen. Does "apt changelog" work for you? | 22:37 |
systemdlete | # apt changelog | 22:37 |
systemdlete | E: Handler silently failed | 22:37 |
systemdlete | :( | 22:37 |
systemdlete | sorry guys | 22:37 |
mason | got to give it a package name | 22:37 |
systemdlete | (be nice if it indicated that... ok, let me try) | 22:37 |
xinomilo | apt changelog linux-image-amd64 | 22:37 |
xinomilo | eg | 22:37 |
systemdlete | I tried apt changelog host (which is one of the pkgs it says is upgradeable) | 22:38 |
systemdlete | E: Failed to fetch changelog:/bind9.changelog Changelog unavailable for bind9=1:9.10.3.dfsg.P4-12.3+deb9u7 | 22:38 |
systemdlete | :( again | 22:38 |
mason | try "apt changelog bash" so you can see what it ought to look like | 22:39 |
systemdlete | Do I need apt-changelog-systemd? | 22:39 |
mason | Nah, it works here, you just need something with a changelog published. | 22:39 |
systemdlete | All I want is the change for the update/upgrade. I don't want the entire history (though that may be useful in other contexts) | 22:40 |
systemdlete | So if a pkg doesn't publish a changelog, it breaks the tool? What am I missing here? | 22:41 |
systemdlete | btw, mason, xinomilo, onefang: I am on ascii where I have this issue. Haven't tried it on beowulf yet. | 22:42 |
systemdlete | (one calamity at a time) | 22:42 |
onefang | I've not tried apt changelog, but apt-listchanges has always worked for me, even if there is- missing changelogs I'm also an ASCII.. | 22:43 |
onefang | And I still need to finish waking up and learn to type. lol | 22:43 |
mason | systemdlete: Ah, only tried it on Beowulf here. I don't have any ASCII lingering. | 22:44 |
user282069 | gnarface: i've rebooted yes. that's a handy idea re: the live image. | 22:44 |
systemdlete | onefang: "apt-listchanges --apt host" gives me the error | 22:47 |
systemdlete | "apt-listchanges host" tells me that the pkg doesn't have a .deb extension | 22:48 |
systemdlete | I wonder if I need a perl module that somehow did not get installed? | 22:48 |
onefang | It works when you do an apt update. | 22:48 |
systemdlete | I've done the update, but I can do it again... | 22:48 |
systemdlete | same | 22:49 |
onefang | If there's nothing to update, there's no changelogs to show. | 22:49 |
systemdlete | apt list --upgradeable shows me a list of a dozen or more updates | 22:51 |
systemdlete | I'm upgrading anyway. | 22:53 |
systemdlete | forget it. I'll try this on beowulf when I have a chance | 22:54 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!