Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
Pablo Bluth
Sep 7, 2007

I've made a huge mistake.
I was only suggesting a simple script to avoid the manual initiation. While rsync is the more advanced program, for this scp should be sufficient and has the massive advance of being a default windows utility.

Adbot
ADBOT LOVES YOU

Agrikk
Oct 17, 2003

Take care with that! We have not fully ascertained its function, and the ticking is accelerating.

Sheep posted:

FSx for Windows (and FSx for Lustre, which isn't being used here) and EFS are not the same thing.

You can mount FSx for Windows shares from Linux, though it seems a little cumbersome. Alternatively some slightly-to-very terrible options are:

1. Use WinSCP and drag and drop the folder. Pretty straightforward and my preferred solution were I doing this exactly once.

This is an hourly batch job.

quote:

2. Install NFS client for Windows and mount the EFS share directly in the Windows machine: it's available in control panel/programs/turn Windows features on.

Amazon EFS is not supported on Windows instances.

quote:

3. Install Samba on the Ubuntu machine and make a share available and reachable from the Windows machine.

Like the above poster suggested. This seems the most promising for my workload.

quote:

4. Install WSL and use rsync.

This is interesting. I've not heard about this.

quote:

If you really want to go the rsync route I have a version I built that doesn't require Cygwin, but frankly it's terrible.



Pablo Bluth posted:

Windows now has OpenSSH shipped by default, so you should have access to scp as a way to copy files. Setup authorized_keys for passwordless access in a script (on a suitably low permission linux account), then write a simple script and schedule it to copy every hour. Use -u or --ignore-existing to avoid recopying old files.

I like this idea. It requires the least configuration of the destination web servers, or the creation of a jump box for the windows share.

RFC2324
Jun 7, 2012

http 418

Does rsync work going to windows? Its more complex than just an scp since it has to runs itself as a reciever instead of just straight writing

Sheep
Jul 24, 2003
Hourly task to copy via scp would be easiest/most straightforward then since OpenSSH stuff seems to be available in Windows now.


Amazon's documentation elsewhere says that EFS uses NFSv4.1 so I think this page may actually be incorrect.

Edit: this should absolutely work on Windows and if not I'd be supremely curious to know the technical reason why:

Sheep fucked around with this message at 17:10 on Aug 15, 2021

BlankSystemDaemon
Mar 13, 2009




Using ssh to natively connect from a windows client to a unix server in powershell will never not strike me as weird.

It does seem like unofficial rsync binaries are a thing but for reasons which should be rather obvious, I can't vouch for them in any way.


EDIT:

Sheep posted:

Hourly task to copy via scp would be easiest/most straightforward then since OpenSSH stuff seems to be available in Windows now.

Amazon's documentation elsewhere says that EFS uses NFSv4.1 so I think this page may actually be incorrect.
I wouldn't touch Amazon EFS without a site-to-site VTI, at least not before NFSv4.3 where hopefully NFS over TLS will have been ratified.
So far though, it only exists in the FreeBSD implementation.

BlankSystemDaemon fucked around with this message at 17:07 on Aug 15, 2021

Agrikk
Oct 17, 2003

Take care with that! We have not fully ascertained its function, and the ticking is accelerating.
From exierience as of yesterday, AWS EFS won't work on windows. Apparently Windows NFS client is implemented slightly differently than AWS EFS. After hours of trying I couldn't make a windows EC2 box see EFS. This was confirmed in a support case I opened up about this very thing.

Hence my need for this ask.

Sheep
Jul 24, 2003
I think I figured it out - the Windows NFS client is only v2/v3, not v4, so it fails out.

Edit: I too built an rsync-without-Cygwin binary but yeah, it's so bad that I deleted the whole project. I'd rather use literally anthing else, including Samba, which I hate touching. Windows' native scp and a scheduled job will be your easiest solution for this project.

Sheep fucked around with this message at 17:55 on Aug 15, 2021

BlankSystemDaemon
Mar 13, 2009




Sheep posted:

I think I figured it out - the Windows NFS client is only v2/v3, not v4, so it fails out.

Edit: I too built an rsync-without-Cygwin binary but yeah, it's so bad that I deleted the whole project. I'd rather use literally anthing else, including Samba, which I hate touching. Windows' native scp and a scheduled job will be your easiest solution for this project.
Yep, the client versions of Windows only support NFS v2 and v3 - so it's a mystery how they test NFSv4.

They don't.

Ffycchi
Jun 4, 2014

Sigh...challenge accepted...shitty photoshop incoming.

Volguus posted:

Share that folder from ubuntu via samba. I have no idea what "Amazon EFS" is, but if it's mounted in the ubuntu servers as a folder somewhere, and they all see the same thing (that pool of ubuntu servers), and presuming that if one writes a file there the others will just see it, then sharing the thing via samba could work. And then it still remains a simple robocopy job as it's just a share as far as windows is concerned.

Oh god please use NFS over samba. Yes, you can use NFS shares on windows.

Also especially for thousands of SMALL files, it's basically built for NFS.

NFS has notably better transfer rates than SMB with small and medium sized files. SMB and NFS are even with large files.

The only things SMB does better is sharing over the internet due to thinks like NAT translation, and an idiot proof setup that requires at most one Braincell. NFS requires a bit more due to ACLs but that's also a boon in terms of granular control.

Also robocopy and rsync are the correct answers outside that.

BlankSystemDaemon
Mar 13, 2009




Ffycchi posted:

The only things SMB does better is sharing over the internet due to thinks like NAT translation, and an idiot proof setup that requires at most one Braincell. NFS requires a bit more due to ACLs but that's also a boon in terms of granular control.
This is one of the major changes with NFSv4, since everything that was implemented using RPC (ie. statistics, locking, mounting, et cetera) was moved into the protocol itself, and everything is now sent over port 2049 (for either TCP or UDP, but UDP is mostly only useful for LAN).
Since everything now happens on one port, it's incredibly easy to create access control lists and overload address translation.

And as NFS over TLS is a thing now (at least in FreeBSD), and will probably be part of v4.3, it basically means there's no reason not to do NFS everywhere.

Worf
Sep 12, 2017

If only Seth would love me like I love him!

I'm a noob so I apologize if this question is framed incorrectly or is partially founded on dumbness

One of the big reasons I was/am excited to be off of windows is I was hoping I would be able to use the s2 sleep state for the PC, which was stolen from me. I'm googling and it sounds like the str/s2ram or whatever is the same thing? I'm not exactly sure which I should be using right now since there seems to be some concepts and terminology I don't yet know.

Anyway, just looking to mimic that function I used to have in windows, do you ppl have some advice about what I should look for? Just want to go back to the days of my laptop suspending but not using a lot of battery.

also I cant seem to find a utility with GUI as good as throttlestop? any recs? if there isnt one with a GUI i would appreciate a recommendation on what I could read to accomplish the same thing. Ive done googling on this as well, which leads me to believe probably not. processor is an 8550u

Agrikk
Oct 17, 2003

Take care with that! We have not fully ascertained its function, and the ticking is accelerating.

Ffycchi posted:

Oh god please use NFS over samba. Yes, you can use NFS shares on windows.

Also especially for thousands of SMALL files, it's basically built for NFS.

NFS has notably better transfer rates than SMB with small and medium sized files. SMB and NFS are even with large files.

The only things SMB does better is sharing over the internet due to thinks like NAT translation, and an idiot proof setup that requires at most one Braincell. NFS requires a bit more due to ACLs but that's also a boon in terms of granular control.

Also robocopy and rsync are the correct answers outside that.


This raises an interesting situation in that I'd now be mounting an Amazon EFS filesystem to an Ubuntu EC2 instance, then sharing that mount point as a new mount point via NFS to a windows EC2 host. Ugh. This sounds kludgy as hell.

Computer viking
May 30, 2011
Now with less breakage.

Pack the linux side into a docker image (or whatever is cool these days), tag it as an NFS version adapter, and pretend that it's a good thing?

NihilCredo
Jun 6, 2011

iram omni possibili modo preme:
plus una illa te diffamabit, quam multæ virtutes commendabunt

Speaking of networked file storage...

I have a Raspberry Pi4 (also running Linux) acting as a 24/7 personal webserver for Nextcloud and other stuff.

Whenever my desktop (running Fedora Silverblue) starts up, which is usually once a day for a couple of hours, I would like it to connect to the Pi (they're on LAN) and make a local differential backup of the Pi's Nextcloud folders.

I already mounted the Pi's storage as a remote folder via the Nautilus built-in SSHFS support, so I thought borgbackup would be right up this alley. I installed Vorta since I'm a noob. However, while I can easily set a local folder as destination, it won't let me choose the remote SFTP folder as a source. I can't pick it in the UI, and if I try and insert the full path manually it says 'Some of your sources are invalid'.

The rejected path looks something like this: /run/user/1000/gvfs/sftp:host=192.168.1.30,port=19876/home/nextcloud

I googled a bit and found this guide which initially sounds exactly what I'm trying to do (backup from a SSHFS mount to a local hard drive) but I can't figure out what or why the guide is doing stuff like... mounting the remote root folder as a chroot? Copying the borg executable on the remote target (and then deleting it)??

It mentions some concerns about UID:GID name mapping but everything I have is single-user so I don't give a poo poo about that, it could all be owned by root for all I care. I just wanted to be able to tell Vorta / Borgbackup to open the same SSHFS connection that Nautilus seamlessly opens on its own, and then treat it the same as a regular folder to be backup it.

Mr Shiny Pants
Nov 12, 2012

NihilCredo posted:

Speaking of networked file storage...

I have a Raspberry Pi4 (also running Linux) acting as a 24/7 personal webserver for Nextcloud and other stuff.

Whenever my desktop (running Fedora Silverblue) starts up, which is usually once a day for a couple of hours, I would like it to connect to the Pi (they're on LAN) and make a local differential backup of the Pi's Nextcloud folders.

I already mounted the Pi's storage as a remote folder via the Nautilus built-in SSHFS support, so I thought borgbackup would be right up this alley. I installed Vorta since I'm a noob. However, while I can easily set a local folder as destination, it won't let me choose the remote SFTP folder as a source. I can't pick it in the UI, and if I try and insert the full path manually it says 'Some of your sources are invalid'.

The rejected path looks something like this: /run/user/1000/gvfs/sftp:host=192.168.1.30,port=19876/home/nextcloud

I googled a bit and found this guide which initially sounds exactly what I'm trying to do (backup from a SSHFS mount to a local hard drive) but I can't figure out what or why the guide is doing stuff like... mounting the remote root folder as a chroot? Copying the borg executable on the remote target (and then deleting it)??

It mentions some concerns about UID:GID name mapping but everything I have is single-user so I don't give a poo poo about that, it could all be owned by root for all I care. I just wanted to be able to tell Vorta / Borgbackup to open the same SSHFS connection that Nautilus seamlessly opens on its own, and then treat it the same as a regular folder to be backup it.

Try mounting it as SSHFS through the CLI instead of the filemanager on a real mount directory. This way it should be treated as a "mount" instead of a Nautilus/Dolphin thingy. I don't know if it changes anything, but this is the way I remotely develop on a Pi filesystem and everything treats it as just a normal directory from the hostmachine.

This way the path should be something like: /mnt/pi.

Mr Shiny Pants fucked around with this message at 07:40 on Aug 17, 2021

BlankSystemDaemon
Mar 13, 2009




If they're on a LAN, is there a reason not to use NFS over SSHFS since it incurs at a whole bunch of context switches by going from the kernel (where the packet comes) to userland (where the ssh process lives) to kernel (to be part of the VFS), and finally to the userland again?

NihilCredo
Jun 6, 2011

iram omni possibili modo preme:
plus una illa te diffamabit, quam multæ virtutes commendabunt

Mr Shiny Pants posted:

Try mounting it as SSHFS through the CLI instead of the filemanager on a real mount directory. This way it should be treated as a "mount" instead of a Nautilus/Dolphin thingy. I don't know if it changes anything, but this is the way I remotely develop on a Pi filesystem and everything treats it as just a normal directory from the hostmachine.

This way the path should be something like: /mnt/pi.

Good point. My only concern is that I had previously used SSHFS on Windows 10 via WSL2 to access the Pi's storage, and it would fail to reconnect automatically if either the desktop or the Pi had rebooted. Hopefully if I stick it as an automount on /etc/fstab it will be more robust.


BlankSystemDaemon posted:

If they're on a LAN, is there a reason not to use NFS over SSHFS since it incurs at a whole bunch of context switches by going from the kernel (where the packet comes) to userland (where the ssh process lives) to kernel (to be part of the VFS), and finally to the userland again?

I'll keep it in mind if the performance is an issue. But the Pi64 is exposed to the internet so I prefer to keep the OS as small as possible to minimize the attack surface - it currently runs SSH, fail2ban, iptables, docker, and nothing else.

BlankSystemDaemon
Mar 13, 2009




NihilCredo posted:

Good point. My only concern is that I had previously used SSHFS on Windows 10 via WSL2 to access the Pi's storage, and it would fail to reconnect automatically if either the desktop or the Pi had rebooted. Hopefully if I stick it as an automount on /etc/fstab it will be more robust.

I'll keep it in mind if the performance is an issue. But the Pi64 is exposed to the internet so I prefer to keep the OS as small as possible to minimize the attack surface - it currently runs SSH, fail2ban, iptables, docker, and nothing else.
Well, sshfs also involves fuse which is a pretty big kettle of fish, and is responsible for the many context switches.
It's also not just a question of performance - the fewer context switches you have, the more time the CPU cores spend idle, and the less power the system uses.

If you configure things properly along with setting up appropriate access control lists on your gateway, I don't see why you can't run NFS.

NihilCredo
Jun 6, 2011

iram omni possibili modo preme:
plus una illa te diffamabit, quam multæ virtutes commendabunt

BlankSystemDaemon posted:

Well, sshfs also involves fuse which is a pretty big kettle of fish, and is responsible for the many context switches.
It's also not just a question of performance - the fewer context switches you have, the more time the CPU cores spend idle, and the less power the system uses.

Yeah, but with sftp or sshfs I don't need to have any extra configuration or running services on the Pi, just the already-existing SSH server. All the 'kettle of fish' stays on the client.

And in a gaming system with a GTX1080 I don't particularly care about saving a few CPU watts for the couple hours a day it stays on :)

quote:

If you configure things properly along with setting up appropriate access control lists on your gateway, I don't see why you can't run NFS.

My gateway is a Huawei-made, ISP-provided residential modem/router combo, I wouldn't trust it to secure a kitten litterbox. If I ever do pull the plug and get a Fritzbox I'll consider offloading some tasks to the router.

Mr Shiny Pants
Nov 12, 2012

NihilCredo posted:

Good point. My only concern is that I had previously used SSHFS on Windows 10 via WSL2 to access the Pi's storage, and it would fail to reconnect automatically if either the desktop or the Pi had rebooted. Hopefully if I stick it as an automount on /etc/fstab it will be more robust.

This is true, I just do a manual mount / unmount.

RFC2324
Jun 7, 2012

http 418

I use borg, and its probably worth mentioning it works best backing up to ssh(not sshfs).

Use this format:

ssh://<user>@<server>:22/<full remote path>

And it should work fine. It was easier to get working that way than it was with a local filesystem

Paul MaudDib
May 3, 2006

TEAM NVIDIA:
FORUM POLICE
Is there a way to get the ubuntu-distro look and feel to actually fully mirror the other distros, like can you import the profile skeletons and stuff so you get a fully populated experience in terms of system utilities and so on?

I did a literal debootstrap to lxde install (following those ZFS on root tutorials) and the UX from just installing LXDE was totally not the same as Lubuntu... but Lubuntu doesn't have ZFS on Root support in the installer.

(I did find a workaround for the graphics problems that led to the debootstrap approach - for some reason you need to loop through the grub loader, back to the UEFI settings, then do a boot override to the grub loader, and add nomodeset. If you loop through the UEFI the iGPU will init properly after nomodeset and you can run the normal install, if you just boot right up and add nomodeset it comes up in 1024x768 and this is broken in ultrawide (buttons are out of sight and can't reach the move/resize bar, and I wasn't good enough to toggle them blind). No, I have no idea why looping through the UEFI would actually work - maybe the Intel UEFI is setting things up before the boot override somehow, and with nomodeset it gets farther into the init before trying it? this is the J5005 atom NUC, NUC7PJYH. It crashes with a DRM error if you boot right up, perhaps gemini lake are different enough to cause problems in the early boot? It works once the system gets installed, and if you loop through the UEFI and add nomodeset it also initializes fine..)

Paul MaudDib fucked around with this message at 08:00 on Aug 21, 2021

Rocko Bonaparte
Mar 12, 2002

Every day is Friday!
I am confused by the question. I believe you can install "stock" versions of the desktop environment; I was leaning on the to try to learn KDE and not Kubuntu. But then you start talking about your boot configuration and I just don't see how that follows everything.

RFC2324
Jun 7, 2012

http 418

Rocko Bonaparte posted:

I am confused by the question. I believe you can install "stock" versions of the desktop environment; I was leaning on the to try to learn KDE and not Kubuntu. But then you start talking about your boot configuration and I just don't see how that follows everything.

yeah, its like he asking if its possible to put a boat build for the atlantic in the pacific.... whats the body of water got to do with anything?

CaptainSarcastic
Jul 6, 2013



My response was more like "I'm glad I abandoned Ubuntu entirely by 2010."

Computer viking
May 30, 2011
Now with less breakage.

If I understand him correctly
- He has to use a clunky debootstrap installer method to get root on ZFS, since that's not yet in the Lubuntu installer.
- If you install Lubuntu, you get a fair bit of customisation that you do not get by just installing LXDE on Ubuntu.
- So: is there a way to get the Lubuntu tweaks on plain Ubuntu?

If I had to guess, there is something like a lubuntu-desktop metapackage you can install - but I don't know if lubuntu has a separate repo for its stuff, or if everything just lives in the normal Ubuntu repos?

Separately, isn't there a plain-ubuntu release that supports root on ZFS that he could start from?

waffle iron
Jan 16, 2004

Computer viking posted:

If I understand him correctly
- He has to use a clunky debootstrap installer method to get root on ZFS, since that's not yet in the Lubuntu installer.
- If you install Lubuntu, you get a fair bit of customisation that you do not get by just installing LXDE on Ubuntu.
- So: is there a way to get the Lubuntu tweaks on plain Ubuntu?

If I had to guess, there is something like a lubuntu-desktop metapackage you can install - but I don't know if lubuntu has a separate repo for its stuff, or if everything just lives in the normal Ubuntu repos?

Separately, isn't there a plain-ubuntu release that supports root on ZFS that he could start from?

Based on 5 minutes looking at the lubuntu website, they haven't put out ISOs for anything since Ubuntu 19.04 base. And looking at Ubuntu package search there is a lubuntu-desktop that indeed pulls in a bunch of lubuntu named packages: https://packages.ubuntu.com/hirsute/lubuntu-desktop

Paul MaudDib
May 3, 2006

TEAM NVIDIA:
FORUM POLICE

Computer viking posted:

If I understand him correctly
- He has to use a clunky debootstrap installer method to get root on ZFS, since that's not yet in the Lubuntu installer.
- If you install Lubuntu, you get a fair bit of customisation that you do not get by just installing LXDE on Ubuntu.
- So: is there a way to get the Lubuntu tweaks on plain Ubuntu?

If I had to guess, there is something like a lubuntu-desktop metapackage you can install - but I don't know if lubuntu has a separate repo for its stuff, or if everything just lives in the normal Ubuntu repos?

Separately, isn't there a plain-ubuntu release that supports root on ZFS that he could start from?

Yeah, exactly.

My problems with the graphical mode were keeping me from doing the normal installer, and the safe mode appears to be broken (at 1024x768 some of the controls are offscreen, and I couldn’t figure it out). I can now do the graphical installer now that I figured out that workaround, and the installer does have a ZFS install option.

I think I was installing the lxde package and not lxde-desktop, I’ll give that a try the next time I’m poking at that machine.

Also Lubuntu has definitely put out 20.04 and 21.04 ISOs.

CaptainSarcastic
Jul 6, 2013



The last edition of Ubuntu I ran was the one before they went to Gnome 3, because gently caress that. At this point it's been ages since I ran a .deb distro of any kind. MEPIS used to be one of my go-tos, and I think it was .deb, but if I recall it's been folded into another project and I lost track of it.

Antigravitas
Dec 8, 2019

Die Rettung fuer die Landwirte:
Does Ubuntu use the task- metapackages that Debian uses? At least in Debian those are what you select in the installer when it asks you what desktop environment you want.

nescience
Jan 24, 2011

h'okay
How does one go about troubleshooting/identifying a possible vulnerability?

My hosting company told me my server is spamming outgoing packets on port 123 (NTP it looks like? I recall NTP being a common service to target for DDOS botnets). I just setup implicit deny on my ufw (minus my SSH whitelist), how should I proceed next to examine what happened? I checked auth.log and nothing seems out of the ordinary at the time of the spam but I'm really not sure what I'm looking for.

Methanar
Sep 26, 2013

by the sex ghost

nescience posted:

How does one go about troubleshooting/identifying a possible vulnerability?

My hosting company told me my server is spamming outgoing packets on port 123 (NTP it looks like? I recall NTP being a common service to target for DDOS botnets). I just setup implicit deny on my ufw (minus my SSH whitelist), how should I proceed next to examine what happened? I checked auth.log and nothing seems out of the ordinary at the time of the spam but I'm really not sure what I'm looking for.

ntp reflection attacks work by the packet requesting the time being smaller than the packet the ntp server responds.

If you send a packet with a fake source IP to an NTP server. the NTP server responds to the fake source IP. for every 20 bytes you send, the NTP server sends like 100. so you get a 5x amplification.

just denying the public internet from hitting your NTP is probably enough.

BlankSystemDaemon
Mar 13, 2009




UDP has an inherit problem in that it's easy to use for reflection attacks.

The reason why NTP was such a big problem is that in addition to being a reflection attack, it's also an amplification attack - with NTP being used to launch record breaking attacks at back in the mid 2010s, which really highlighted how big the problem can be, and caused US-CERT to issue a warning about potential future attack vectors based on amplification ratios.

Paul MaudDib
May 3, 2006

TEAM NVIDIA:
FORUM POLICE
OK, I can confirm that installing the lubuntu-desktop metapackage on the mainline ubuntu installer doesn't get you the same thing as installing Lubuntu. There's a bunch of UI configuration and packages that aren't installed.

the approach I've been trying to do is install ubuntu on the main drive, install Lubuntu on a separate drive, boot the lubuntu install, mount the zfs volumes, delete the main install and rsync the lubuntu copy in. So far I haven't quite got that to work right though.

I guess maybe the easy option is dumping a list of packages and installing that inside ubuntu, and then copying out the user skeleton and any files it depends on, and then putting those in Ubuntu?

nescience
Jan 24, 2011

h'okay

Methanar posted:

ntp reflection attacks work by the packet requesting the time being smaller than the packet the ntp server responds.

If you send a packet with a fake source IP to an NTP server. the NTP server responds to the fake source IP. for every 20 bytes you send, the NTP server sends like 100. so you get a 5x amplification.

just denying the public internet from hitting your NTP is probably enough.

Ok thanks. Just a proper ufw setup then.

Buck Turgidson
Feb 6, 2011

𓀬𓀠𓀟𓀡𓀢𓀣𓀤𓀥𓀞𓀬

Paul MaudDib posted:

OK, I can confirm that installing the lubuntu-desktop metapackage on the mainline ubuntu installer doesn't get you the same thing as installing Lubuntu. There's a bunch of UI configuration and packages that aren't installed.

the approach I've been trying to do is install ubuntu on the main drive, install Lubuntu on a separate drive, boot the lubuntu install, mount the zfs volumes, delete the main install and rsync the lubuntu copy in. So far I haven't quite got that to work right though.

I guess maybe the easy option is dumping a list of packages and installing that inside ubuntu, and then copying out the user skeleton and any files it depends on, and then putting those in Ubuntu?

Holy moly surely there's just some dot files you can copy over

RFC2324
Jun 7, 2012

http 418

Buck Turgidson posted:

Holy moly surely there's just some dot files you can copy over

I mean, thats half of it, just copying over the homedir, but lubuntu probably does have some packages not pulled in by the ubuntu metapackage(tho as I understand this isnt supposed to be true in theory) and even more probably has some snowflake stuff configured in /etc or /usr/lib or some other place that isn't easy to catch.

I would honestly ask on the lubuntu forums, thats a question thats extremely distro specific, and I feel like most of our ldxe users are off the ubuntu family at this point

xzzy
Mar 5, 2009

I've got a (in my mind) simple udev rule that is causing kernel panics on boot:

code:
IMPORT{program}="/sbin/blkid -o udev -p %N"
ACTION=="add", KERNEL=="sd[a-z][0-9]", ENV{ID_FS_LABEL}=="EOS_DIGITAL", SYMLINK+="MediaSource"
During boot, the kernel panic comes from the blkid command.

code:
[    9.259906] Unable to handle kernel paging request at virtual address fee00000
[    9.267265] pgd = (ptrval)
[    9.270033] [fee00000] *pgd=80000000007003, *pmd=00000000
[    9.275549] Internal error: Oops: 206 [#1] SMP ARM
[    9.280422] Modules linked in: brcmfmac brcmutil sha256_generic cfg80211 v3d gpu_sched rfkill raspberrypi_hwmon i2c_bcm2835 bcm2835_codec(C) bcm2835_isp(C) v4l2_mem2mem bcm2835_v4l2(C) videobuf2_dma_contig bcm2835_mmal_vchiq(C) videobuf2_vmalloc videobuf2_memops videobuf2_v4l2 videobuf2_common spi_bcm2835 videodev mc vc_sm_cma(C) vc4 cec drm_kms_helper rpivid_mem drm drm_panel_orientation_quirks snd_soc_core snd_compress snd_pcm_dmaengine snd_pcm snd_timer nvmem_rmem snd syscopyarea sysfillrect sysimgblt fb_sys_fops backlight uio_pdrv_genirq uio i2c_dev ip_tables x_tables ipv6
[    9.332744] CPU: 1 PID: 717 Comm: blkid Tainted: G         C        5.10.52-v7l+ #1441
This is on a raspberry pi running on a fully updated install of raspbian. If I remove the udev rule the system boots clean. If I get the system running, install the rule, reload udev's config and plug the media in everything works as expected. But on the next boot it'll all go south.

Can anyone spot if I'm doing something the wrong way?

Takes No Damage
Nov 20, 2004

The most merciful thing in the world, I think, is the inability of the human mind to correlate all its contents. We live on a placid island of ignorance in the midst of black seas of infinity, and it was not meant that we should voyage far.


Grimey Drawer
I put together a FreeNAS box and have a bunch of app jails, all of which is based on FreeBSD. One day I noticed that I could filter my command history by typing in some characters before hitting the up arrow and it would skip back through my previous commands but only show me ones that matched those characters. What is this 'filtering' called and can I get terminals from other distros to do the same thing? I can't explain why but I really love being able to do that and miss it whenever I'm CLI'ing around on my Linux machines.

Adbot
ADBOT LOVES YOU

Methanar
Sep 26, 2013

by the sex ghost

Takes No Damage posted:

I put together a FreeNAS box and have a bunch of app jails, all of which is based on FreeBSD. One day I noticed that I could filter my command history by typing in some characters before hitting the up arrow and it would skip back through my previous commands but only show me ones that matched those characters. What is this 'filtering' called and can I get terminals from other distros to do the same thing? I can't explain why but I really love being able to do that and miss it whenever I'm CLI'ing around on my Linux machines.

reverse history search
ctrl-R

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply