Arm-dev August 2018

arm-dev@lists.centos.org

17 participants
22 discussions

Boot Raspberry Pi image in KVM on CentOS 7
by Brenton Earl 22 Aug '18

22 Aug '18

Hello, I am trying to boot the Raspberry Pi image in qemu KVM on CentOS7. I converted the image to a qcow2 file system, extracted the /boot directory and tried to boot. I am using the following command: $ sudo virt-install --name centos7_armhfp \ --memory 4096 \ --boot kernel=/var/lib/libvirt/armhfp-boot/boot/vmlinuz-4.14.28- 201.el7.centos.armv7hl,initrd=/var/lib/libvirt/armhfp- boot/boot/initramfs-4.14.28- 201.el7.centos.armv7hl.img,kernel_args="console=ttyAMA0 rw root=/dev/sda3" \ --disk /var/lib/libvirt/images/CentOS-Userland-7-armv7hl-Generic- Minimal-1708-guest.qcow2 \ --import \ --arch armv7l \ --machine virt \ --os-variant centos7.0 I get the following error: Starting install... ERROR unsupported configuration: qemu-xhci not supported in this QEMU binary Domain installation does not appear to have been successful. If it was, you can restart your domain by running: virsh --connect qemu:///system start centos7_armhfp otherwise, please restart your installation. I am following this tutorial that should technically work on all architectures if modified to fit: https://arrfab.net/posts/2017/Sep/29/using-centos-7-armhfp-vm-on-centos -7-aarch64/ What am I doing wrong? I searched the web a bit but found no significant results. I tried to boot with the generic and the pi images with the same error output. I feel like I am missing something simple because I am used to using virt-manager instead of the command line to install virtual machines. Thanks, -- Brenton Earl <brent(a)exitstatusone.com>

3 3

Re: [Arm-dev] New kernel takes ~130MB!
by Pablo Sebastián Greco 16 Aug '18

16 Aug '18

El 16/8/18 a las 15:57, Robert Moskowitz escribió: > Fine, but does yum delete the old before installing new or afterwards? No, it is done after a successful installation, as a part of the "cleanup" stage. > > BTW, in Fedora 28 we had so many problems with the kernels, that many > of had to run 6 kernels back. Finally seem to have the problems > addressed... The think is that it always keeps the current booted kernel, so if you are using a "working" kernel, it shouldn't be a problem. > > On 08/16/2018 02:50 PM, Pablo Sebastián Greco wrote: >> >> I always change installonly_limit in /etc/yum.conf from the default 5 >> to 2, so you only keep the current and the new one. >> >> El 16/8/18 a las 15:44, Robert Moskowitz escribió: >>> I just installed the new kernel on one of my Cubieboard2s. /boot >>> used grew ~130MB. >>> >>> Challenge is that my next board to update only has 43MB free. It >>> currently has 3 kernels on it. >>> >>> Is there a way to cleanly delete old kernel files prior to the >>> update? In this case 4.9.30-203? >>> >>> thanks >>> >>> >>> _______________________________________________ >>> Arm-dev mailing list >>> Arm-dev(a)centos.org >>> https://lists.centos.org/mailman/listinfo/arm-dev >

1 0

Security auditing service
by Robert Moskowitz 07 Aug '18

07 Aug '18

I had to reboot a server this morning. I was working on one box and bumped things I should not have... Then I noticed my name server was not coming up real fast like normal. So I get on the console and notice it is waiting for something to start. It waited 3 min before giving up and getting everything else going. So after it was up, I ssh in and checked the messages and found: Dec 31 19:00:29 onlo systemd: Starting Security Auditing Service... Dec 31 19:00:29 onlo auditd[499]: Started dispatcher: /sbin/audispd pid: 501 Dec 31 19:00:29 onlo audispd: No plugins found, exiting Dec 31 19:01:59 onlo systemd: auditd.service start operation timed out. Terminat ing. Dec 31 19:03:30 onlo systemd: auditd.service stop-final-sigterm timed out. Killi ng. Dec 31 19:03:30 onlo systemd: auditd.service: control process exited, code=kille d status=9 Dec 31 19:03:30 onlo systemd: Failed to start Security Auditing Service. Dec 31 19:03:30 onlo systemd: Unit auditd.service entered failed state. Dec 31 19:03:30 onlo systemd: auditd.service failed. Ignore the timestamp. This occurs before the network is up and chrony has not set the time. Cubies do not have a battery RTC. But I believe this is recent. I had gone a LONG time without updating and applied some 180 rpms updates (in careful batches). So this MAY be something new. systemctl -l status auditd Shows ● auditd.service - Security Auditing Service Loaded: loaded (/usr/lib/systemd/system/auditd.service; enabled; vendor prese t: enabled) Active: failed (Result: signal) since Wed 1969-12-31 19:03:30 EST; 48 years 7 months ago Docs: man:auditd(8) https://github.com/linux-audit/audit-documentation Process: 498 ExecStart=/sbin/auditd (code=killed, signal=KILL) Dec 31 19:00:29 onlo.htt-consult.com systemd[1]: Starting Security Auditing Serv ice... Dec 31 19:00:29 onlo.htt-consult.com auditd[499]: Started dispatcher: /sbin/audi spd pid: 501 Dec 31 19:01:59 onlo.htt-consult.com systemd[1]: auditd.service start operation timed out. Terminating. Dec 31 19:03:30 onlo.htt-consult.com systemd[1]: auditd.service stop-final-sigte rm timed out. Killing. Dec 31 19:03:30 onlo.htt-consult.com systemd[1]: auditd.service: control process exited, code=killed status=9 Dec 31 19:03:30 onlo.htt-consult.com systemd[1]: Failed to start Security Auditi ng Service. Dec 31 19:03:30 onlo.htt-consult.com systemd[1]: Unit auditd.service entered fai led state. Dec 31 19:03:30 onlo.htt-consult.com systemd[1]: auditd.service failed. So what may be going on? My web server, medon.htt-consult.com is practically identical and is not exhibiting this behavior. Dec 31 19:00:30 medon auditd[499]: Started dispatcher: /sbin/audispd pid: 501 Dec 31 19:00:30 medon auditd[499]: Init complete, auditd 2.8.1 listening for events (startup state enable)

2 2

Security Audit service failing
by Robert Moskowitz 07 Aug '18

07 Aug '18

This is on a Cubieboard2 I took the CentOS-Userland-7-armv7hl-generic-Minimal-1804-sda image, did yum install policycoreutils-python passwd hostnamectl set-hostname nevia.htt-consult.com adduser -c "Robert Moskowitz" -G wheel rgm passwd rgm timedatectl set-timezone America/Detroit vi /etc/ssh/sshd_config # change port number semanage port -a -t ssh_port_t -p tcp nnn systemctl restart sshd.service firewall-cmd --permanent --add-port=nnn/tcp firewall-cmd --permanent --remove-service=ssh firewall-cmd --reload Yum update poweroff power on, and on the console see messages: [ OK ] Started Tell Plymouth To Write Out Runtime Data. [ OK ] Started Import network configuration from initramfs. Starting Create Volatile Files and Directories... [ OK ] Started Create Volatile Files and Directories. Starting Security Auditing Service... [ *** ] A start job is running for Security...ting Service (3min 3s / 3min 8s) Then it takes off. I could scan the console capture if you need the messages immediately after it finally took off. But I have noticed this service failing to start on one of my other boxes, not not on the other two. So it revolves around something installed or not installed. But with this base of an install, perhaps you can figure out what the problem is? thanks

1 1

RPI / kernel config
by Stephan Guilloux 06 Aug '18

06 Aug '18

Hi all, Is there any way to know the compilation options used to build the kernel ? At least, on a x86_64, we have something like config-3.10.0-862.3.3.el7.x86_64 On raspberry, I did not find anything in /proc nor in /boot. Regards, Stephan.

3 5

Re: [Arm-dev] Cubieboard5 performance
by Patrick Laimbock 06 Aug '18

06 Aug '18

Hi Bob, Pablo, -----Original message----- > >> Finally, the main thing about that board (about that SoC actually), > >> is that SATA is not native, it is an integrated SATA-USB, so > >> performance may be an issue. > >> > > > > Ouch. this is for a mail server that has extensive disk I/O. So > > forget this board. > > > >> Personally I love the BananaPi M2Ultra, 2G RAM, 4 Cores and native > >> SATA. The only problem is that SATA support is not integrated yet in > >> any kernel, so it needs external patches. > > Of course if there is a quad core 64-bit ARM with sata under $100 that > > runs at 5V; that would do nicely! It costs more than $100 and I'm not sure if it's 5V or 12V but the new LattePanda boards might be interesting: https://www.kickstarter.com/projects/139108638/lattepanda-alpha-soul-of-a-m… BR, Patrick

2 1

What is ifcfg-link?
by Robert Moskowitz 03 Aug '18

03 Aug '18

I went looking for /etc/sysconfig/network-scripts/ifcfg-eth0 to set it up for static addressing. Instead I found ifcfg-link What is this? In the past I was usin nmcli: ifname=eth0 nmcli con delete $ifname nmcli con add type ethernet con-name $ifname ifname $ifname ip4 $your_ipv4_address/$your_ipv4_prefix gw4 $your_ipv4_gateway nmcli con mod $ifname ipv4.dns "$your_ipv4_dns1 $your_ipv4_dns2" # optionally set your MAC address # your_mac= # nmcli con mod $ifname mac "$your_mac" nmcli con up $ifname (that con add above is line wrapping in this email) ip a shows eth0; do I just use my nmcli script as I did for the 1611 image? thanks

1 3

Machiatobin network interfaces
by Skorpeo Skorpeo 03 Aug '18

03 Aug '18

Hi, I was able to install centos on the machiatobin using the iso via efi. However no network interfaces show up, I also tried to plug in a usb wifi adapter, it is recognized in dmesg but still no network interface with command ip a (only shows loopback device) or cat /proc/net/dev. Any pointers greatly appreciated. I also tried going into bios and selecting ACPI setting in the device manager menu but no change... Also, as an aside, to be able to install using the iso I had to add kernel parameter console=ttyS0,115200n8 on boot as per this post: http://macchiatobin.net/forums/topic/feedback-on-uefi-boot-with-mini-iso/. Wasn't sure if it would be helpful to document that. Incidentally, that poster also saw no network interfaces. Thanks much.

2 1

update-boot?
by Robert Moskowitz 02 Aug '18

02 Aug '18

The wiki says: Before centos-userland-release-7-5.1804, in order to activate the new kernel and edit /boot/extlinux/extlinux.conf, you had to run /usr/bin/update-boot. This is no longer valid and it is done automatically by grubby. But I just happened to look at /boot/grub/grub.conf default=0 timeout=5 splashimage=(hd0,1)/grub/splash.xpm.gz hiddenmenu title CentOS-Userland-7-armv7hl-generic-GNOME-1804 (4.14.28-201.el7.centos.armv7hl) root (hd0,1) kernel /vmlinuz-4.14.28-201.el7.centos.armv7hl ro root=UUID=be2c9311-e289-4cab-b878-89b522144d70 initrd /initramfs-4.14.28-201.el7.centos.armv7hl.img And I did the yum update that installed the 4.14.52 kernel. Should I run update-boot? /boot/klist.txt has: 4.14.28-201.el7.centos.armv7hl 4.14.52-201.el7.armv7hl

2 4

Request for the Xfce desktop
by Robert Moskowitz 01 Aug '18

01 Aug '18

I am a proponent of Xfce, especially for armv7 boards. I got introduced to Xfce with Fedora-arm 19 (or was it 18 or even 17?). I did the Gnome image install, dropped to runlevel 3, installed vncserver, and do get the gnome desktop with my notebook taking most of the performance hit. But having at least the option of installing the minimal image then doing a group install "Xfce Desktop" would be great. BTW, I did 'yum list >y.lst' and 'grep -i xfce y.lst' and came up empty handed. Is Xfce even built for arm? thanks

4 7

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

Arm-dev August 2018