freenode/#devuan/ Monday, 2019-05-06

glats&window close03:32
glatsups03:32
DeeEffHey all03:38
DeeEffFirefox has had a bad weekend03:38
DeeEffwondering if ceres is going to see their new build anytime soon03:39
fsmithredit'll probably be a couple weeks before new mini isos can be made03:40
DeeEffspecifically version 66.0.4 (right now ceres is on 66.0.1) :(03:40
fsmithredceres?03:40
fsmithredoh, you're talking about ff versions03:40
DeeEffI mean, I'm on ceres, and wondering if I can apt-upgrade that package to 66.0.4 sometime soon03:40
fsmithreddepends on debian's schedule on that.03:41
DeeEffah, so we just forward that package from debian?03:41
fsmithredyou can always get it from mozilla03:41
fsmithredyeah03:41
DeeEffwell, I'd rather not have a local install, I like being able to upgrade whenever I can, and also so I don't forget later03:42
fsmithredlike 95% of the packages03:42
Unit193In the meantime you can just use Mozilla's hotfix, no?03:42
fsmithredyeah, it works03:42
fsmithredI'm using it03:42
fsmithrednoscript is still working03:42
DeeEffNo, the default firefox install in debian disables their tracking stuff by default03:43
DeeEffthe studies box is greyed out03:43
DeeEffI can't turn it on.03:43
fsmithredxpinstall.signatures.required;false03:43
fsmithredabout:config ^^^03:44
DeeEffoh well dang03:44
DeeEffguess I need to set a reminder to re-enable this in like... two to three weeks?03:45
DeeEffthanks03:45
DeeEffI thought that was only in beta versions03:45
fsmithredyeah, good idea03:45
DeeEffnot stable03:45
furrywolfit's come out yesterday that disabling studies in the config page, and how debian does it, firefox still phones home and does whatever mozilla tells it to do.  specifically, their fix will still be applied even if on the debian build with studies disabled.03:45
Unit193fsmithred: That's not the hotfix, the hotfix is https://storage.googleapis.com/moz-fx-normandy-prod-addons/extensions/hotfix-update-xpi-intermediate%40mozilla.com-1.0.2-signed.xpi  That's just a workaround that disables verification.03:45
DeeEfffurrywolf: well yes, their fix is orthogonal to studies. They used studies as a quick way to deploy code.03:46
furrywolf"normandy" is still enabled even if you try to disable studies and the like, which phones home to mozilla with a bunch of information, and lets mozilla edit the browser's config (and other things).03:46
DeeEffUnit193: that's not the hotfix, but it does get me going until Debian updates their firefox package in sid03:46
DeeEff¯\_(ツ)_/¯03:47
Unit193DeeEff: Why do you say it isn't the hotfix?  FWIW, they've been uploaded, just need to build and dinstall.03:47
DeeEffsorry, I meant the xpinstall thing isn't the hotfix (agreeing with you)03:47
DeeEffbut I also don't want to build / install a separate version of firefox, I'd rather use the package on apt.03:47
Unit193Ah!  I understand.  The hotfix I linked is just an xpi/addon.03:48
furrywolfI wonder how many other ways firefox is phoning home even after you try disabling everything...03:48
furrywolfI should file a bug report with debian that they're disabling studies but not disabling normandy.03:53
Unit193https://bugs.debian.org/92843303:53
furrywolfsomeone beat me to it.  :)03:54
furrywolfthe best part is mozilla didn't tell anyone they did this, and it slipped through everyone's auditing.03:54
XenguyMozilla was alright taking a beating, and now because of this they're taking another beating04:18
XenguyIt seems to me FF just got hijacked by suits somewhere along the line04:20
XenguyIt certainly doesn't feel much like the pluckish browser that came along at the time and said we're out to build the best FOSS web browser ever04:22
XenguyFor awhile, they were great04:24
XenguyBut now, alas04:25
slvrUnit193: Did that bugfix close reason make any sense to you? I don't understand.05:57
Unit193slvr: That the option in question is implicitly disabled by app.shield.optoutstudies.enabled and/or browser.onboarding.shieldstudy.enabled being false.06:06
DocScrutinizer05I'm pretty late on this one... nevertheless: using CSS and/or HTML in email may cause massive risk to compromise crypto privacy/signature https://arxiv.org/ftp/arxiv/papers/1904/1904.07550.pdf  https://github.com/RUB-NDS/Johnny-You-Are-Fired08:36
clemens3morning, just upgraded from jessie to ascii11:49
clemens3worked and rebooted11:49
clemens3but now apt-get update gives hash mismatch error11:50
clemens3hash sum mismatch.. tried country name de.deb.devuan.org in sources.list11:50
xinomiloprobably mirror sync issue, try again in a while or switch mirror11:55
clemens3hmm, ok11:55
Evilhamhellooo11:58
Evilhamclemens3: around?11:58
clemens3re12:03
clemens3Evilham: yes?12:03
Evilhamhi hi12:03
Evilhamso, question time!12:03
Evilhamwhere are you located?12:03
Evilhamare you using ascii?12:04
clemens3Switzerland12:04
clemens3i dist-upgrade to ascii from jessie12:04
clemens3that seemed fine, then rebooted12:04
clemens3used deb.devuan.org12:04
EvilhamI see, yes, I managed to identify the issue now with my computer (as opposed to before at 4am with my phone :-p)12:05
clemens3cool12:05
Evilhamthat's the good news, the bad news is that we are now waiting for replies :-p-12:05
clemens3ok, problem identified.. then I am happy and wish good fixing.. waiting mode..12:06
Evilhamin any case, if you are in CH, you should totally use https://mirror.ungleich.ch/mirror/packages/devuan/merged as the base for your repo :-D they have a pretty good connection speed12:06
clemens3i will, before I wasnt sure how to prefix.. now you showed me..12:06
Evilhamyeah, ideally you wouldn't have had to :-)12:07
clemens3I adapted my sources.list to the mirror12:10
clemens3you give an update when the situation change?12:11
clemens3d12:11
Evilhamsure thing, I can ping you here when that happens12:11
clemens3super, thanks a lot!later12:12
cosurgihi, did the devs abandom devuan, or everything will be okay? I love devuan, please don't kil it due to aprils' joke.12:32
rrqEvilham: I've had 'apt-get update' issue for 2 hours now. even from ungleich :(12:34
fsmithredcosurgi, devuan is not dead. Everything will be ok.12:34
EvilhamReally? Shouldn't be the case12:37
Evilhamrrq: will you be around in about 30 mins?12:37
rrqsure12:37
EvilhamGood12:38
cosurgifsmithred: thanks!13:06
xinomiloused devuan tor mirror, new firefox with fix, is in.14:29
clemens3any update? or is there a workaround? I have a production server in limbo..14:36
Evilhamclemens3: it's fixed15:55
Evilhampropagating now15:55
Evilhamif you need it very urgently, use pkgmaster.devuan.org15:55
Evilhamwithin the next couple hours everything should be up-to-date15:56
fsmithredEvilham, I must have started my upgrade two seconds after it was ready, because it just completed successfully.15:58
Evilhamit's been fixed for shy over 30 mins now15:59
fsmithredah, ok.16:01
clemens3ok, thanks , will check now..16:05
clemens3error is gone, super.. thanks a lot!16:11
Evilhamthank you for helping with the debugging :-p16:14
fsmithredxinomilo, what version of firefox-esr do you have now?16:18
slvrI asked a while ago about packaging icecat for devuan. I'm willing to do the work but I don't know the build system for the repos. any hints?16:47
clemens3as some feedback.. I had set a trap for myself, there was some age old setup to prevent systemd on a pre jessie debian setup.. still living in the background.. my fault, then openssh-server update was not done properly and myeself locked out.. but could only find it after the apt-get updated worked again..16:49
fsmithredslvr, you could start with importing the source to git.devuan.org16:53
fsmithredbest if you can clone an existing repo that has the history16:53
fsmithredclone/fork I'm not sure which16:54
fsmithredget it working and people can test it16:55
Evilhamfix is (almost) fully propagated \o/16:55
fsmithredcool, thanks16:55
slvrfsmithred: the git tree from gnu only contains a script to pull firefox from mercurial, modify, then build it.17:05
fsmithredslvr, that might work. I know of one other package we pulled from mercurial17:06
slvrI would think the newly generated icecat code would be what needs to be checked in17:06
fsmithredyeah, that would be best17:06
fsmithredyou can't find that?17:07
slvrI have it on my workstation, but I'd think we would generate a new one to package.17:07
slvrmaybe modify all the sed and file move operations to be tracked in git.17:08
fsmithredthe newest source package I'm seeing is from november17:12
fsmithredis there anything newer?17:12
slvrno, ff 60 esr is current17:14
slvrlooks like a new ff esr from mozilla is due in july17:15
fsmithredhave you built it locally yet?17:16
slvrI have. Currently working on reproducing that build on a fresh system.17:17
Evilhamthe last mirror that was out of sync finished about 1h ago20:54
Evilhamso, deb.devuan.org is fully back to normal20:55
g4570n👍21:31
flrnHello!21:47
flrnI hope for some hints on my ascii desktop stalling at boot for some (varying) seconds when loading the "initial ramdisk".21:50
flrnthis phenomenon is quite new (some weeks)21:51
flrnI am not aware of any changes besides the usual updating21:51
flrnno SSD involved, legacy BIOS (this machine does not have UEFI)21:52
flrnI am a bit afraid that it might be a weak capacitor?!21:53
gnarfaceflrn: check in /etc/default/grub to see if you still have "quiet" in the kernel command-line options.  if so, remove it, then you should more accurately be able to see what is stalling21:54
flrngnarface: yes it's quieted, thanks I'll fix and reboot...21:55
flrngnarface: the reboot went blazingly fast now21:58
gnarfaceflrn: well, there should be a lot more text so there would be an illusion of it being faster even if it's not21:58
flrnwill do a full shutdown now, to test if it happens only on a fully discharged system21:59
gnarfaceyea a cold boot might make a difference21:59
gnarfacebut it could be a networking issue too21:59
flrnnetwork at this stage?21:59
gnarfacesure.  if your /etc/hosts doesn't properly define localhost and at some point one of those regular updates pulled in a MTA you didn't have before (which would be expected behavior in many cases) then a simple delay in response from your DNS or DHCP server (typically your plastic toy home router) could trigger up to a 60 second timeout22:01
gnarfacethat's just an example22:01
gnarfacebut i had you remove "quiet" in the hope that we could be sure22:01
gnarfacesince whatever is stalling, usually happens right after the last line of text before the hang22:02
gnarfaceif it's not reliably reproducible still, unfortunately that is not helpful information.  you gotta first find a way to trigger the delay reliably.  if the harddrive is going bad enough to trigger noticable startup delays, i'd expect smartmontools to be able to show you some other early warning signs of failure22:04
gnarfaceif nothing, absolutely nothing in the install or the hardware itself seems to be possibly the culprit, but you are using DHCP, then the DHCP server is a very likely culprit22:05
gnarface(statistically followed by DNS configuration, if those two things aren't coming from the same place, but note that /etc/hosts properly set up should mitigate both issues)22:06
gnarfaceusually bad capacitors cause stability issues, not performance issues22:07
furrywolfif only we had a more modern init system that would do everything in parallel so it wouldn't stall...  :P22:07
* furrywolf hides22:07
blebanyone notice that keyboard shortcuts dont work in libreoffice impress?22:08
gnarfacefurrywolf: it does actually do everything in parallel that it can, but a bunch of network daemons, once installed, are considered dependencies of other stuff, and can hang on network configuration or connectivity issues.  (this is a big reason i strongly favor static network configurations, actually)22:08
blebthis is a relatively fresh ascii install, just trying ctrl+z and it's not undoing, but selecting it from the menu works22:08
gnarfacei never touched it bleb, sorry, can't say.22:09
gnarfaceit's only in impress?  not the other libreoffice suit tools?22:09
gnarfacesuite*22:09
furrywolfbleb:  I've never used impress.22:10
gnarfacebleb: i do notice there's libreoffice in ascii-backports, just fyi.22:10
flrngnarface: oh, MTA triggers something... but the system does not hang noticeable anymore since I removed the "quiet" option22:13
gnarfaceflrn: that's really weird.  all i can advise is to keep observing it to see if you notice anything new the next time there's a delay.22:15
gnarfacea long time ago (pre-4.x kernels) i had a strange boot bug where the output text would halt until i pressed a key22:16
gnarfaceit wasn't supposed to be doing that22:16
gnarfaceand all the king's horses and all the king's men couldn't figure out what was up22:17
gnarfacei later traced it to unexpected behavior from some bios powersaving setting that wasn't working right22:17
gnarfacethat's a really rare type of issue though.  usually intermittent startup delays are caused by the network.22:18
gnarfaceso what i'm saying basically is that it's highly unlikely that just removing "quiet" would fix the startup delay in and of itself.   i wouldn't rule it out completely, but i would leave that conclusion for the very last.22:20
gnarfaceonce, i put a modern wpa2 enabled wifi dongle in a old P-II laptop and discovered that the longest the DHCP server would wait was within microseconds of the fastest the machine could respond (running wpasupplicant) which would lead to a long delay followed by failure to negotiate a network connection, just about 9 out of 10 tries22:26
furrywolfheh, my irc client often doesn't notice the network has dropped until I press a key.22:26
gnarfacethen once it got it, it would cache the response somehow, and it wouldn't be a problem again unless the laptop was powered off for a few hours22:26
gnarfacebut in the end it seemed to be unique to the DHCP server (another plastic toy issued by the ISP) and not something that could be adjusted with any supplied configuration options at either end22:28
furrywolfyeah, that seems like a pretty broken dhcp server22:28
gnarfaceyea they're supposed to wait 60s too, but it clearly wasn't doing that.22:29
gnarfacethe issue was hard to diagnose by reading forum reports about it though, because people with faster machines were not experiencing the failure nearly as often22:29
gnarfacein fact, it took me a real long time to figure out that it would work if i just kept trying22:30
furrywolfI probably would have just jumped to "this dhcp server is broken, use something else" without tracking down why.22:32
flrngnarface: the delay happens long before any NICs are brought up.22:35
gnarfaceflrn: interesting, though that may actually be the cause, if something is coming up early that needs the network.  does it always stall on the same line of text?22:36
gnarfaceflrn: also, is it a laptop?22:37
gnarface(laptop bioses might have powersaving features enabled for spinning-platter harddrives that can cause some delay while they "wake up")22:38
furrywolfheh, how about a feature to preheat hard drives before booting?  :)22:39
gnarfacelike an oil pan heater on a car?22:39
furrywolfyep22:39
gnarfaceheh, might actually be useful in some environments...22:39
furrywolfmy toughbooks have hard drive heaters...  I've never turned that feature on, however.22:39
gnarfacefor real? i didn't know that was a thing22:40
flrnwhen I "unquiet" grub, there is no noticeable delay. "quiet" the delay happens right after the "loading initial ramdisk" line.22:40
flrnbut I got an old error message back: ata[1-4] soft reset failed. there are no p-ata disks in that system and IIRC it is even disabled in the BIOS.22:41
furrywolfgnarface:  https://d3inagkmqs1m6q.cloudfront.net/2188/media-photos/pcdd2613-genuine-hard-drive-caddy-tray-panasonic-toughbook-cf-19-free2-3day-ship-mk1-mk6-2.jpeg  it's a resistive heater element sheet that wraps around the drive22:42
gnarfaceflrn: *that* is interesting.  if you're sure that is the case, you could try just blacklisting that driver in the kernel22:42
flrn"it" == the p-ata disks22:42
gnarfaceflrn: (i'd make sure you had a workable live image around just in case you're wrong though, so you have a way to turn it back on)22:43
furrywolfif you turn it on, it'll pre-heat the drive if it's below a certain temperature before spinning it up22:43
gnarfacefurrywolf: cool22:43
furrywolf"might actually be useful in some environments" describes many toughbook features.  :)22:44
flrngnarface: I have some grub-bootable isos on the boot partition -_- but IIUC, I'd need to do the blacklisting for the initramfs, no?22:45
flrnor is it really sufficient to create a blacklist.conf under /etc/modprobe.d?22:47
flrnit seems to be sufficient, https://wiki.debian.org/KernelModuleBlacklisting states:22:48
flrn1. Create a file '/etc/modprobe.d/<modulename>.conf' containing 'blacklist <modulename>'. / 2. Run 'depmod -ae' as root / 3. Recreate your initrd with 'update-initramfs -u'22:49
flrngnarface: I blacklisted ata_generic and pata_atiixp, but no change: still failing ata soft reset messages.22:59
flrnnot sure, what else to blacklist22:59
flrnlibata.force= works somewhat different than expected... but it has an effect (trying to recover now)23:33
gnarfaceflrn: sorry, stepped away for a bit there, but i have nothing to add now except that maybe you're right about bad capacitors after all.  if there's a bad cap causing some intermittent failure of some ata bus handshake or something like that... i don't know for sure that it wouldn't behave like this.  anyway, in theory disabling it in the bios *and* the kernel SHOULD quiet it.23:49
gnarfaceusually on a motherboard the sockets and the busses themselves are the last things to go, but i've still seen it happen.23:50
gnarfacethis type of behavior is more common in my experience with aging usb devices23:51
gnarfacebut theoretically i think any type of bus connection could fail in this way23:51
gnarface(dmesg exposing periodic reset warnings could be the smoking gun here)23:52
gnarfacei'd google around for the warning messages and see if anyone else has seen them for this particular disk controller23:53
gnarfacei'd also search to figure out for sure which kernel modules it's using to make sure they're the ones you blacklisted23:53
gnarfacethere are a dozen or two of of them23:54
gnarfaceand some overlap in compatibility23:54
gnarface(i.e., it's not supposed to happen anymore, but it isn't unheard of for it to load the wrong one after you've blacklisted the right one)23:54
flrngnarface: I had to find out, that ata[1-4] are my sata devices, not the unused p-ata channels. I have also updated the jessie-netinst on my boot partition to ascii now -_-23:56
gnarfaceah i see.  that was a possibility that occurred to me.23:57
gnarfacei have no way to conclude whether it is a hardware or software issue23:58
gnarfaceif you disable unused SATA ports, it might still improve boot time, but i wouldn't really expect the delays related to that to occur after grub23:59

Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!