|
I recommend at least giving ChromeOS with crouton a shot before you start from scratch. It's quite nice. Unless you're doing something that you can't do in a chroot I suppose.
|
# ? Sep 11, 2014 21:00 |
|
|
# ? May 27, 2024 02:51 |
|
If you mostly just use Chrome anyways, I'd second sticking with Chrome OS and using crouton to setup a chroot for stuff that Chrome OS doesn't natively support.keyvin posted:I like mine, but I have a lot more time than I have money so I am more than willing to futz. The other neat thing about Chromebooks is that they're arguably the most open hardware on the market. The source code to most of the firmware is available from the Chromium OS git repos. It even includes firmware source for the embedded controller, which is very proprietary on all other devices. Granted, making firmware modifications may be a pain in rear end and have little practical benefit, but it's refreshing to see a company go in that direction with their hardware (or rather, force OEMs to go in that direction).
|
# ? Sep 12, 2014 00:11 |
|
ExcessBLarg! posted:So which other laptop runs Linux flawlessly? Yeah if you get a sufficiently old ThinkPad it will work with most modern desktop distributions. But I definitely had to futz around a lot back in the day to get my T61 working, and had to futz a lot with it again a few months ago when Debian switched to systemd and broke all my poo poo. Perhaps I have just had extremely good luck over the years, but I would find it difficult to think of a laptop that didn't work flawlessly under Linux. The only sticking point that is coming to mind at the moment is 5ghz wireless adapters, but I am used to just running wireless G anyway. Also, I guess I am assuming that you would be running proprietary drivers and such, and not insisting on a completely open source environment. My little Acer AO722 netbook runs vastly better under Linux than it does under Windows, including with more up-to-date graphics drivers and such.
|
# ? Sep 12, 2014 00:59 |
|
ExcessBLarg! posted:So which other laptop runs Linux flawlessly? Yeah if you get a sufficiently old ThinkPad it will work with most modern desktop distributions. But I definitely had to futz around a lot back in the day to get my T61 working, and had to futz a lot with it again a few months ago when Debian switched to systemd and broke all my poo poo. Up until the c720 I bought two random sub $400 laptops and everything worked right out of the box. The c7 and c720 is the first time I had to futz with *anything* since ~2007. Really since X started detecting modelines automagically and wifi manufacturers split the firmware out of their drivers I haven't had trouble installing Linux on other people's laptops either. I mean, yeah if I tried to install Debian stable on a newish laptop, it probably wouldn't be fun, but... For the people running Crouton, did you change the cboot flags? Is there a non warranty voiding way to keep from losing everything when someone fucks up and hits space?
|
# ? Sep 12, 2014 01:56 |
|
CaptainSarcastic posted:Perhaps I have just had extremely good luck over the years, but I would find it difficult to think of a laptop that didn't work flawlessly under Linux. Graphics have been all over the map: Intel, despite long being opensource had some weird stuff in the i945 days, like having to patch the VESA BIOS to set native resolution. Nvidia is a choice between reverse-engineered drivers that support X.org extensions well but overheat and might not do 3D vs. the proprietary driver which doesn't support as many X extesnsions (e.g., xrandr) and had weird performance problems with Firefox whenever I used it. Oh, and the thing where the wrong Nvidia card would take five minutes to boot and beep a lot. I haven't used ATI in a long time, but when I did, the fglrx drivers were unstable as poo poo and it seems it took them a long time to figure that out (if they ever did). Hotkeys, backlight adjustment, and touchpad might or might not work. ACPI S-states might fail, corrupt data, etc. Perhaps things have been better in recent years. The last non-Chromebook laptop I purchased was a ThinkPad 410i in 2010.
|
# ? Sep 12, 2014 04:39 |
|
keyvin posted:For the people running Crouton, did you change the cboot flags? Is there a non warranty voiding way to keep from losing everything when someone fucks up and hits space? Not sure, I haven't looked into it since I turned on developer mode. I don't do anything that isn't synced up somewhere else, whether through git/hg or Dropbox, and I save all the changes I make to in crouton as shell scripts/package lists/etc., so I'm not too concerned. Though I can see why this would be a dealbreaker for some people. I'd be interested to hear if there's a workaround. A few times I've hit space by accident and it does give you a confirmation and a chance to back out.
|
# ? Sep 12, 2014 05:23 |
|
The linux-chromebook modified kernel for Arch distros is pretty much a drop in replacement that fixes any issues on the Acer C720 and other Chromebooks. With it installed you don't need to do any of the parts in the Arch C720 wiki pages involving the Touchpad, Touchscreen, Suspend or Kernel parameters. Just remember to mask out the tmp directory ($ systemctl mask tmp.mount) if you compile the kernel on the Chromebook itself or you'll run out of RAM. Unmask it once the new kernel is installed. Its also unnecessary to remove any screws from the C720 to use it with linux. After you enable developer mode in ChromeOS and reboot, you hit CTRL-L to load the SeaBios which will boot from a thumbdrive if it can or the hard drive. It does mean you'll need to hit CTRL-L anytime you reboot but you shouldn't have to do that much. This also prevents you from ever bricking the device. I've been using my C720 with Arch and Openbox for months now. Its pretty good for coding and browsing and gets insane battery life. The screen is a bit crap but its a $200 dollar laptop, can't expect perfection. Sauer fucked around with this message at 06:50 on Sep 12, 2014 |
# ? Sep 12, 2014 06:41 |
|
ExcessBLarg! posted:Wireless has been an on and off again pain. Yeah, I should've made a more general statement about that - I've seen enough stuff about it on Linux forums that it should have occurred to me. When I was doing more installs I had access to a variety of wireless cards that I could swap in and out until I found one that worked, so was a little spoiled. I really didn't start putting Linux on laptops until 2008-2009, and was working in a shop with lots of hardware I could change out as needed. Rather than troubleshooting too deep on any given distro, I would usually just throw distros at a machine until one stuck. Ubuntu, MEPIS, and PCLOS were the ones I used most, and generally one of those three would stick (with some degree of fiddling). From 2011 or so (whenever the 12.2 release was) I have been installing pretty much nothing but OpenSUSE, as it has been just dandy on everything that I've thrown it at.
|
# ? Sep 12, 2014 07:04 |
|
ExcessBLarg! posted:Nvidia is a choice between reverse-engineered drivers that support X.org extensions well but overheat and might not do 3D vs. the proprietary driver which doesn't support as many X extesnsions (e.g., xrandr) and had weird performance problems with Firefox whenever I used it. Just in case that makes any difference, Nvidia started shipping rudimentary XRandR support (up to 1.2, I think), which is good enough for screen setup etc., a while ago.
|
# ? Sep 12, 2014 13:49 |
|
Sauer posted:Its also unnecessary to remove any screws from the C720 to use it with linux. After you enable developer mode in ChromeOS and reboot, you hit CTRL-L to load the SeaBios which will boot from a thumbdrive if it can or the hard drive. It does mean you'll need to hit CTRL-L anytime you reboot but you shouldn't have to do that much. This also prevents you from ever bricking the device. The issue isn't using it with Linux, the issue is accidentally taking it out of developer mode. Maybe they updated what happens when you hit space, but my brother accidentally did it. I had to re-install chromeOS on the SSD, then re-install Linux. It was after that I said screw it and opened it up so I could permanently change the boot behavior. Like I said in my post, you *can* leave chromeOS on it so you don't have to go through that. The downside is you lose a lot of space on a small SSD, and you have to alter the partition table in a precise way or chromeOS won't be able to boot anyways.
|
# ? Sep 12, 2014 13:55 |
|
CaptainSarcastic posted:I really didn't start putting Linux on laptops until 2008-2009, and was working in a shop with lots of hardware I could change out as needed. It's entirely possible that, after all this time, hardware has been sufficiently commoditized that most models really do just work out of the box. Although things like the T440 UEFI brick bug still give major pause--it's actually that exact problem that convinced me to investigate alternatives. Anyways I freely admit that moving to Chrome-branded hardware is trading one set of frustrations for another. I still find it refreshing though. Here's the thing: I wouldn't recommend that someone purchase a Chromebook or Chromebox unless they actually wanted to run Chrome OS on it. Sure, the Intel models can run regular Linux desktop distributions, and in the case of laptops, Chromebooks are decent hardware for the price. For small form-factor desktops I'd purchase an Intel NUC instead. By running Chrome OS in developer mode and installing a chroot environment with crouton, you can do pretty much all desktop Linux activities in a reasonably convenient way. There are still frustrations though, perhaps the biggest once is the monthly Chrome OS force update which has a tendency to break developer mode behavior in subtle ways. So I still spend time futzing with stuff. I guess what makes it worthwhile for myself is that Chrome OS, even in verified mode, handles 90% of my daily computing tasks out of the box. I do use dev mode shells and occasionally drop into an X session out of convenience, but it's only required for about 10% of what I do. Thus, it's incredibly convenient being able to grab random Chrome hardware, login and sync, and be off on my work. When my past Linux laptops died, I'd be unproductive for a day or more.
|
# ? Sep 12, 2014 14:18 |
|
keyvin posted:The issue isn't using it with Linux, the issue is accidentally taking it out of developer mode. The bigger problem is if you want to run Chrome OS in dev mode without accidentally blowing away the contents of the RW partition. I don't know, here's it's just not been a problem in practice as I reboot my Chromebook rarely. I do keep important/bulk data on an SD card (ext4 is natively supported, LUKS can be used with some effort). So worst case I have to rerun dev_install and crouton, but I don't lose anything critical.
|
# ? Sep 12, 2014 14:25 |
|
ExcessBLarg! posted:The last time I tried a full Ubuntu install I blew away all the partitions on the SSD. They're not required for legacy boot. After doing that, pressing spacebar required that I insert restore media since there no longer existed enough of the verifiable partitions for the device to restore verified mode without it. I'm fairly certain that after doing so (pressing spacebar, rebooting at the insert media screen) left Ubuntu intact. You aren't "pretty" sure. I think you are drat sure . There is no way you could forget having to re-install chromeOS. After my brother hit space, that was it. It booted directly to the ChromeOS is missing or damaged screen with no way to re-enable developer mode w/o re-isntalling chromeOS. I wonder if there are multiple firmware versions?
|
# ? Sep 12, 2014 15:19 |
|
Yeah there's a crapload of firmware versions. A high level, new devices (Bay Trail?) are transitioning from Coreboot + U-Boot to Coreboot + Depthcharge. On a device-specific basis, part of the firmware is ready-only and can't be field upgraded. That might actually include all of the recovery process since it's needed to restore from bad flashes, I'm not sure. I'm fuzzy on the exact behavior because I did restore Chrome OS after installing Ubuntu. I can't recall now if I did so only intentionally, or once accidentally too. You might be right. Edit: And it was on a Chromebox. The C720 being one of the first Haswell device is almost certainly running an older firmware tree. ExcessBLarg! fucked around with this message at 15:54 on Sep 12, 2014 |
# ? Sep 12, 2014 15:52 |
|
Oh, if anyone does buy a c720, make sure to claim your goodies before wiping chromeOS. Mine came with 100gb of google drive for two years.
|
# ? Sep 12, 2014 16:05 |
|
If it helps re:different firmware, the Chromebook on which I've occasionally hit space and been saved by a confirmation screen is a C7 (C700?) from a couple years ago.
|
# ? Sep 12, 2014 17:01 |
|
I'm trying to install Arch Linux and I'm stuck on this command "gummiboot install" getting this error message "Failed to create EFI Boot variable entry: No space left on device". I did a bit of googling and that error seems to be related to nvram being full. I did try deleting some unused entries using efibootmgr with the command code:
The ones that are left: code:
I'm trying to dual boot it with Windows 8 since there's still some games that aren't Linux compatible yet. JoeMB fucked around with this message at 21:40 on Sep 12, 2014 |
# ? Sep 12, 2014 21:31 |
|
JoeMB posted:I'm trying to install Arch Linux and I'm stuck on this command "gummiboot install" getting this error message "Failed to create EFI Boot variable entry: No space left on device". I did a bit of googling and that error seems to be related to nvram being full. Answer: stop using obscure bootloaders and use grub2 "No space left on device" is probably the EFI partition (FAT, probably the first partition on your device). But it could be you not mounting the efi firmware or some other thing. If the only place you can find people talking about it is the Arch forums, it's probably an unstable pile of crap that nobody should use.
|
# ? Sep 12, 2014 21:48 |
|
What am I doing wrong? I'm trying to setup nginx to reverse proxy various apps running on various ports, for personal use only. I've got http basic auth over SSL setup, and various location blocks. Each one redirects properly, but then as soon as it tries to access some other resource it can't find it. The first app I tried to configure was taskwarrior-web (did this one first because it should be relatively simple). It serves from host:5678/tasks and host:5678/projects. At first I wanted to put it at /task, then I realized it might be easier to put it at /tasks. So I also made a location block for /projects. So I could get to the barebones site but I was getting errors for its css, js, etc. My solution was to find where taskwarrior-web installed itself and symlimk everyting into /usr/share/nginx/html. It works! On to the second app, RStudio server. It doesn't use a subdirectory for its main page so I used the rewrite directive to get rid of the /rstudio/ where I wanted to serve it from. The first thing the server does is redirect to host:port/auth_sign_in. So I made another location block. Then I had the same issue as before with the resources. I found where it keeps everything. But it has a ton of files, so before I started symlinking I figured there must be a better way, and I came here. What am I doing wrong?
|
# ? Sep 13, 2014 00:36 |
|
Ah, that's annoying--the static assets and perhaps even links that the apps generate might be absolute instead of relative. I.e. they're tying to go to http://<your server>:<port>/foo instead of http://<your server>:<port>/application/foo. Looking at taskrabbit it looks like they have a bug on supporting something like that but it's still open. :/ https://github.com/theunraveler/taskwarrior-web/issues/56
|
# ? Sep 13, 2014 00:43 |
|
So what you're saying is usually this sort of thing would be configured on the application rather than the server? E: and there's nothing fundamentally incorrect about adding a nginx location block for every subdirectory and a symlink for every resource an app uses? SurgicalOntologist fucked around with this message at 05:25 on Sep 13, 2014 |
# ? Sep 13, 2014 05:17 |
|
Yeah if it was a nice app it would let you tell it that you're putting it under a specific path and then generate URLs from there. What you're doing makes sense though, I think it's the only real workaround without changing the app code.
|
# ? Sep 13, 2014 06:05 |
|
If that application is the only thing on that vhost, you can just remap /(.*) to /application/\1.
|
# ? Sep 13, 2014 06:20 |
|
evol262 posted:
EFI Boot Variables do not exist on the EFI partition: they are purely in NVRAM. Of course, the error message might be worded with less than 100.0% precision. Also, the efibootmgr command does not display _all_ the EFI variables, just the boot-related ones. Look at /sys/firmware/efi/vars or /sys/firmware/efi/efivars to see the full list. To see them, you may need to have kernel module(s) "efivars" and/or "efivarfs" loaded, and the efivarfs pseudo-filesystem mounted at /sys/firmware/efi/efivars. If you see variables named like "dump-type0-*", you may have been using a kernel that has dumped its "dmesg" output to NVRAM at crash/sleep/shutdown. This was enabled by default when the CONFIG_EFI_VARS_PSTORE kernel feature was first introduced, until it turned out that some buggy EFI BIOSes react badly to NVRAM getting full. (See kernel configuration options CONFIG_EFI_VARS_PSTORE and CONFIG_EFI_VARS_PSTORE_DEFAULT_DISABLE. A nice feature in theory, but in practice the current state of the art of EFI firmware programming is such that you'd only want to activate it after careful consideration or when troubleshooting a specific problem that can't be caught otherwise.) One way to (try to) delete EFI variables from the OS is: code:
Don't delete any variables whose purpose you don't recognize, unless you like finding potentially hardware-bricking firmware bugs.
|
# ? Sep 13, 2014 09:53 |
|
I have a machine that dual-boots Ubuntu and Windows 8.1. However, the Grub 2 menu had an old Windows 7 entry, probably off of my storage disk. I tried to delete it by reading this and this, but I somehow ended up doubling my menu entries, with the second copies starting after all the first copies (i.e. Entry1, Entry2, Entry3, Entry1A, Entry2A, Entry3A rather than Entry1, Entry1A, etc.). Here's what's in /etc/grub.d:code:
code:
code:
|
# ? Sep 13, 2014 15:25 |
|
SurgicalOntologist posted:What am I doing wrong? I'm trying to setup nginx to reverse proxy various apps running on various ports, for personal use only. I've got http basic auth over SSL setup, and various location blocks. Each one redirects properly, but then as soon as it tries to access some other resource it can't find it. Does the try_files directive allow an opportunity to avoid this kind of symlinking? If so I can't quite figure it out.
|
# ? Sep 13, 2014 16:29 |
|
hooah posted:Here's what's in /etc/grub.d: Let's condense that grub.cfg a bit. It is produced mostly by the scripts in /etc/grub.d, and the parts produced by each script is marked with the ### BEGIN <scriptname> and ### END <scriptname>. In the beginning there is the "do not edit this file" comment block, created by the main grub-mkconfig. (edit: some lines split with \ to minimize table breakage) code:
The duplication of your boot menu is because you've apparently filled /etc/grub.d/40_custom with your complete grub.conf file, minus the Windows 7 boot entry. It doesn't work that way unless you disable all the other scripts in /etc/grub.d... and if you do that, you'll have to do some manual work each time your Linux kernel package is updated, so don't do that. First, fix the doubling by editing your /etc/grub.d/40_custom. It should contain a header like this: code:
Then to your ghost-of-Windows-7 issue: The "safe" workaround would be to disable the os-prober (by adding GRUB_DISABLE_OS_PROBER=true" to /etc/default/grub, and then make sure the /etc/grub.d/40_custom contains only the Windows 8 boot entry and not the whole GRUB configuration. The real fix would be to instead make /dev/sda1 (or whatever your ex-Windows7 disk now is) unrecognizable as a Windows 7 boot disk. At least on Debian, the os-prober is a collection of shell scripts, and the script responsible for detecting Windows 7 is /usr/lib/os-probes/mounted/20microsoft. Apparently it looks for the presence of /boot/bcd in the root of the Windows boot disk, so you'll just need to delete or rename that from your ex-Windows7 disk. If you do this, be very careful not to damage your Windows 8 bootloader instead. (It is a hidden directory, so you won't see it in Windows if your Folder Options are on default settings. Or just mount the filesystem on Linux and use your rootly might to remove the remains of the Windows 7 bootloader.) After making the changes, run "sudo update-grub". It will run grub-mkconfig and place the new configuration file in the proper location. telcoM fucked around with this message at 18:22 on Sep 13, 2014 |
# ? Sep 13, 2014 18:17 |
|
SurgicalOntologist posted:What am I doing wrong? I'm trying to setup nginx to reverse proxy various apps running on various ports, for personal use only. I've got http basic auth over SSL setup, and various location blocks. Each one redirects properly, but then as soon as it tries to access some other resource it can't find it. A basic reverse proxy only passes the queries to whatever application and answers back to the browser. It won't edit the content produced by the application in any way. Any serious web application should have some way of configuring what kind of URLs it uses to point to itself, particularly the host part. You should adjust these settings to take your reverse proxy into account. SurgicalOntologist posted:The first app I tried to configure was taskwarrior-web (did this one first because it should be relatively simple). It serves from host:5678/tasks and host:5678/projects. At first I wanted to put it at /task, then I realized it might be easier to put it at /tasks. So I also made a location block for /projects. So I could get to the barebones site but I was getting errors for its css, js, etc. My solution was to find where taskwarrior-web installed itself and symlimk everyting into /usr/share/nginx/html. It works! So... you're essentially reverse-proxying just the main HTML content and letting nginx provide all the resources of taskwarrior-web directly. Yeah, that can actually improve performance on static content over reverse-proxying the whole application, since a web server like nginx is supposed to be very, very good at serving static content. You'll just have to make sure you've caught *all* the non-static stuff with location blocks, or you might see... "unexpected behavior". And if the application engine is vulnerable to some attack using a badly-formed URI, then this might actually give you some extra protection: if the attack URI does not match any of your application-specific Location blocks, it won't get passed to the application at all. Instead, nginx will just reply with a 404. SurgicalOntologist posted:On to the second app, RStudio server. It doesn't use a subdirectory for its main page so I used the rewrite directive to get rid of the /rstudio/ where I wanted to serve it from. The first thing the server does is redirect to host:port/auth_sign_in. So I made another location block. Then I had the same issue as before with the resources. I found where it keeps everything. But it has a ton of files, so before I started symlinking I figured there must be a better way, and I came here. You might want to see if you can configure RStudio to use a specific host:port part (or even host:port/prefix) when generating URLs to point to itself. This is not the same thing as setting where RStudio actually listens for incoming connections! Then have nginx reverse-proxy that to the actual RStudio port. That way you can manage overall access control with nginx, which is probably at least part of the reason why you're reverse-proxying in the first place. These kind of configurations are sometimes needed when applications are integrated into a larger whole or served in a parallel fashion using a load balancer, so it would be semi-reasonable to expect any major Web applications to have the settings needed for that. If this is the configuration for the kind of RStudio you're setting up: http://rstudio.github.io/shiny-server/latest/#default-configuration ...then, changing the code:
code:
|
# ? Sep 13, 2014 19:16 |
|
That's really helpful, thanks. I think I'm starting to get a handle on this stuff. As far as I can tell, RStudio Server doesn't have any options besides where it listens: https://support.rstudio.com/hc/en-us/articles/200552316-Configuring-the-Server Unless I pay $10K for the pro version: https://s3.amazonaws.com/rstudio-server/rstudio-server-pro-0.98.501-admin-guide.pdf What you linked is actually the config for Shiny Server, which is for deploying R applications--RStudio Server is a browser-based IDE (a really cool idea, wish I could run Sublime Text like that). I managed to get it working with four location blocks. Unfortunately I can't disable its built-in authentication without the Pro version, so this whole thing was a bit pointless, after all typing <server>/rstudio/ is not really any more convenient than <server>:<port>, if easier to remember. And actually, my intention all along has been to host it at <server>/rstudio/ as you suggest. But as far as I can tell this makes everything harder, not easier. I guess that wouldn't be the case if there was a prefix setting on the app side. E: It's sort of strange though, as the docs suggest this: code:
code:
Hmm, this suggests something about the headers. Guess I should play around some more. SurgicalOntologist fucked around with this message at 20:50 on Sep 13, 2014 |
# ? Sep 13, 2014 20:39 |
|
evol262 posted:You may want to read through this and try the suggestions. This was ages ago but I forgot to say thanks for the help. I finally had a chance to read through that thread and try a few things and wifi seems more stable now.
|
# ? Sep 14, 2014 17:25 |
|
telcoM posted:Lots of stuff Thanks, I think that mostly helped sort it out. I was at least able to get rid of most of the duplicates. still have a second Windows 8 option. I got rid of the 41_custom script in /etc/grub.d. Here's what I have now: /boot/grub.cfg: code:
code:
hooah fucked around with this message at 20:24 on Sep 14, 2014 |
# ? Sep 14, 2014 19:38 |
|
You can remove the Windows 8 entry from 40_custom, it's still getting added automatically by 30_os-prober.
|
# ? Sep 14, 2014 23:37 |
|
As mentioned, I have a chromebook. I also have a desktop that draws 70 watts at idle, compared to the chromebook's ~10-15. I would like to put some things (such as /var/spool/mail) on an nfs share so I can use the c720 more (with monitor/keyboard/mouse). I have a 3tb drive in an external enclosure. I have tried a raspi as a cheap way to get the external drive on the network. The raspi would frequently lock up under sustained reads and writes even when using a powered hub. Has anyone tried a beagle bone black for this? If the BBB isn't suited, are there any other suggestions?
|
# ? Sep 15, 2014 01:40 |
|
keyvin posted:As mentioned, I have a chromebook. I also have a desktop that draws 70 watts at idle, compared to the chromebook's ~10-15. I would like to put some things (such as /var/spool/mail) on an nfs share so I can use the c720 more (with monitor/keyboard/mouse). I have a 3tb drive in an external enclosure. I have tried a raspi as a cheap way to get the external drive on the network. The raspi would frequently lock up under sustained reads and writes even when using a powered hub. Has anyone tried a beagle bone black for this? If the BBB isn't suited, are there any other suggestions? Anything is better than a pi. Cubie if you want native sata. Beagle if you want to do embedded programming or use BSD. Odroid if you just want fast and don't care about open or hackable
|
# ? Sep 15, 2014 02:05 |
Tried to do a dist upgrade from Xubuntu 12 to 14, now when it boots I get "The disk drive for /dev/mapper/cryptswap1 is not ready yet or not present. Continue to wait or Press S to skip mounting or M for manual recovery". Googling around I see some suggestions for things to try, but my system completely freezes up when this error appears, so I can't really try any of the fixes. What should I do?
|
|
# ? Sep 15, 2014 03:14 |
|
fletcher posted:Tried to do a dist upgrade from Xubuntu 12 to 14, now when it boots I get "The disk drive for /dev/mapper/cryptswap1 is not ready yet or not present. Continue to wait or Press S to skip mounting or M for manual recovery". Googling around I see some suggestions for things to try, but my system completely freezes up when this error appears, so I can't really try any of the fixes. What should I do? Make a bootable linux usb and mount your mdcrypt volume manually so you can fix the issue.
|
# ? Sep 15, 2014 03:34 |
Longinus00 posted:Make a bootable linux usb and mount your mdcrypt volume manually so you can fix the issue. Unfortunately the only external media I have available is a 512MB thumb drive, seems to limit my choice of live cd. I downloaded the Universal USB Installer from pendrivelinux.com and the Lucid Puppy iso. Tried to boot it and got "lupu_528.sfs not found" - but that file is on the thumb drive. Same error when trying to boot from "puppy pmedia=usbflash". Tried drat Small Linux and got a similar error. fletcher fucked around with this message at 05:03 on Sep 15, 2014 |
|
# ? Sep 15, 2014 04:37 |
Alright I got a bigger thumb drive and successfully made my Ubuntu usb boot stick. I'm not really sure where to go from here though.code:
code:
code:
code:
edit: I noticed when I press the power button, the blank screen goes back to the console login prompt for a moment before shutting down. I tried resetting my xorg.conf but that didn't seem to help. fletcher fucked around with this message at 09:51 on Sep 16, 2014 |
|
# ? Sep 16, 2014 05:28 |
|
fletcher posted:I tried simply commenting out the cryptswap1 line from fstab to see if I could get it to boot but it still froze at the Xubuntu splash screen. So I changed "quiet splash" in the boot options to nomodeset and now I just get a black screen when booting. Sounds like "froze at the Xubuntu splash screen" is something like "the X server is not starting properly, leaving the boot splash on screen." In such a situation, removing the "quiet splash" boot options is a good thing: it should give you more information on the boot sequence. However, depending on your display controller/GPU, adding "nomodeset" might be a bad thing: some newer GPUs can be used with a modesetting driver only. I think this includes (most?) Intel iGPUs. According to your /etc/crypttab, your /dev/sda5 should contain a "disposable" encrypted swap area. Its encryption key is initialized from /dev/urandom, so anything in it is effectively lost as soon as the OS is shut down. Because of this, your boot-up scripts should be running a "mkswap /dev/mapper/cryptswap1" sometime after the disk encryption is activated but before swap areas are activated (i.e. after "cryptdisks_start" or similar, but before "swapon"). If that does not happen, your previously-used swap area (when looked through a changed encryption key) does not look like a swap area, but like a partition of random noise.
|
# ? Sep 16, 2014 16:10 |
|
|
# ? May 27, 2024 02:51 |
|
spoon0042 posted:You can remove the Windows 8 entry from 40_custom, it's still getting added automatically by 30_os-prober. That did it. Thanks.
|
# ? Sep 16, 2014 16:11 |