Discuss November 2018

discuss@lists.centos.org

96 participants
69 discussions

CentOS 7 package not present in Red Hat EL 7 - qt-assistant
by Toralf Lund 28 Nov '18

28 Nov '18

I'm using CentOS 7 for development of software that is sometimes used on Red Hat Enterprise Linux 7. I conjunction with an update of one of the applications, I asked some Red Hat users to install the Qt 4 Assistant application via the qt-assistant package (which is used by a "help" function in our software.) It seemed like there was no such package in the Red Hat package set, however, and I also see no mention of it in "package manifests" like https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/7/ht…. Yet I can install the package from the "base" repository on my CentOS machine. Questions: Isn't CentOS base supposed to contain exactly the same packages as Red Hat Enterprise, except in some special cases that relate to distribution information, installation sources etc.? Does anyone know what's going on with the specific package I mention above? - Toralf

3 4

aarch64 boot up process/EFI
by Asya Dvorkin 28 Nov '18

28 Nov '18

Hello, I am not sure if this is the correct place to ask, but I do not have anyone else to go to. I'm working on building our own custom aarch64 RHEL distribution and using CentOS SRPMs to rebuild necessary packages. I am also utilizing kickstart. At this point, my kickstart installs all packages (by using a local repository with newly rebuild packages), but in the end I get an error "failed to write boot loader configuration". I noticed that i have /boot/efi/EFI/redhat directory on my newly "installed" server, while centos server has /boot/efi/EFI/centos directory. According to efibootmgr it is trying to boot \EFI\centos\shimaa64.efi, which doesn't exist (EFI\redhat\shimaa64.efi does). Those files are created by shim rpms and they are a total mess. I don't even know where to start or what to look for. `efibootmgr -v` shows 0000 in the beginning of BootOrder Boot0000* CentOS: HD (1,GPT,.......)/File(\EFI\centos\shimaa64.efi). Shim package has the following files/directories: $ rpm -qlp shim-aa64-12-1 /boot/efi/EFI/BOOT/BOOTAA64.EFI /boot/efi/EFI/BOOT/fbaa64.efi /boot/efi/EFI/redhat/BOOTAA64.CSV /boot/efi/EFI/redhat/mmaa64.efi /boot/efi/EFI/redhat/shim.efi /boot/efi/EFI/redhat/shimaa64-redhat.efi /boot/efi/EFI/redhat/shimaa64.efi I have been reading EFI documentation and I understand the concept (or rather I think I do), but not sure how to make it all work with shim RPMs and/or how to properly rebuild them. Would really appreciate any advice... Thank you! Asya

1 0

RHEL 8 Public Beta Released
by Yan Li 27 Nov '18

27 Nov '18

https://www.redhat.com/en/blog/powering-its-future-while-preserving-present… -- Yan Li

7 13

Re: [CentOS] NBDE, clevis and tang for non-root disk
by mark 27 Nov '18

27 Nov '18

Radu Radutiu wrote: > On Tue, Nov 27, 2018 at 3:14 PM mark <m.roth(a)5-cent.us> wrote: > >> What we do is to have the encryption key of the secondary filesystem in >> /etc/crypttab, which is, of course, 600. As it boots, it decrypts from >> that as it mounts the rest of the system. >> > Thanks, this is working as expected and it gave me the hint needed to > find the actual problem. The problem is that the initramfs image generated > by dracut -f does not include the /etc/crypttab from the OS (it only > contains the entry for the root device). Once I have manually added the > other volumes in the /etc/crypttab file from the initramfs image, clevis > is able to decrypt all volumes. Now the question is why the generated > iniramfs image has a different /etc/crypttab. How can I specify > /etc/crypttab for the initramfs so that > furhter kernel updates will not replace it with the wrong file? > Sorry, I think you misunderstood. The key for root is *not* in /etc/crypttab - that's only for the secondary ones. mark

2 1

C7 install, "failed to IDENTIFY"
by mark 27 Nov '18

27 Nov '18

The install from a USB key fails. It's showing ata2:0.0 failed to IDENTIFY. I've been searching online, and the only hint I have is that it might not understand the controller. New Dell Optiplex 7050. Haanyone run into this? mark

2 1

Re: [CentOS] NBDE, clevis and tang for non-root disk
by Radu Radutiu 27 Nov '18

27 Nov '18

On Tue, Nov 27, 2018 at 3:14 PM mark <m.roth(a)5-cent.us> wrote: > What we do is to have the encryption key of the secondary filesystem in > /etc/crypttab, which is, of course, 600. As it boots, it decrypts from > that as > it mounts the rest of the system. > > mark > Thanks, this is working as expected and it gave me the hint needed to find the actual problem. The problem is that the initramfs image generated by dracut -f does not include the /etc/crypttab from the OS (it only contains the entry for the root device). Once I have manually added the other volumes in the /etc/crypttab file from the initramfs image, clevis is able to decrypt all volumes. Now the question is why the generated iniramfs image has a different /etc/crypttab. How can I specify /etc/crypttab for the initramfs so that furhter kernel updates will not replace it with the wrong file? Radu

1 0

CentOS 7 bash perhaps off topic
by Jerry Geis 26 Nov '18

26 Nov '18

hi all, I have a small script that seems to be exiting and hitting the sleep 60... The smwebsocket just connects to the web socket provided and outputs the data. This works manually. the myprogram just opens a database and writes the line... My desire is to run the smwebsocket, connect to the websocket and output the data (line by line) when we get a line that matches Location take that line and output to the database. Seems simple. I desire this to keep the connection alive and just continue to read data and grep on the data etc... if smwebsocket does exit, sleep 60 and reconnect. However it seems to be exiting and running the sleep 60. Am I missing something ? while [ 1 ] do smwebsocket "$URL" | grep Location | while read line do myprogram -data "$line" done sleep 60 Reconnecting... done All the pieces work - just not the actual running. What am I missing? Jerry

3 3

kernel 3.10.0-957.el7.x86_64 + EFI on Dell server - problem
by lejeczek 26 Nov '18

26 Nov '18

hi guys I've one box where I just yesterday upgraded Centos. I wonder if that kernel upgrade process might somewhat troublesome. After that upgrade UEFI boot fails with: Failed to set MokListRT: Invalid Parameter Something has gone seriously wrong: import_mok_state() failed : Invalid Parameter Has anybody seen that? And maybe know to fix it? many thanks, L.

6 9

rpmbuild errors
by Alessandro Baggi 26 Nov '18

26 Nov '18

Hi list, I need a more updated version of bacula because I need to backup a debian workstation. So I decided to build my own bacula package base on bacula 9.2.0. I'm new to rpmbuild and I created my bacula.spec file but when running rpmbuild -ba bacula.spec I got error about check-rpaths like: ERROR 0001: file '/usr/sbin/bregex' contains a standard rpath '/usr/lib64' in [/usr/lib64] Running rpmbuild as suggested, adding QA_RPATHS=$[ 0x0001|0x0010 ] works without problem but I don't understand what is wrong with my spec files. Another issue that I found is when installing this rpm. It reports that my bacula package requires libbaccats-9.2.0.so()(64bit) but this file is inside rpm so to solve this I must add to my spec file: Provides: libbaccats-9.2.0.so()(64bit) That permit to install my new rpm. How I can solve this 2 issues? Thanks in advance.

1 0

NBDE, clevis and tang for non-root disk
by Radu Radutiu 26 Nov '18

26 Nov '18

Hi, Has anybody managed to get network disk bound disk encryption to work with a non-root disk? It works fine for the root device, but the moment I add another volume to /etc/crypttab the system will no longer boot automatically. A tcpdump on the tang server shows no traffic while the system is stuck at the LUKS password prompt. The second encrypted volume is set up in the same way as the root device and I can unlock the volume using clevis-luks-unlock -d /dev/vda3. I've seen in https://rhelblog.redhat.com/2018/04/13/an-easier-way-to-manage-disk-decrypt… that clevis-luks-askpass.path needs to be enabled but it doesn't make a difference. Any ideas on what 's wrong or how to debug this? Best regards, Radu

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Discuss November 2018