This reverts commit eef9b2529c.
It appears that htpasswd is not salting any passwords that are
stored with the SHA (-s) algorithm. MD5 passwords however are
salted.
That leads us to the conclusion that the "MD5 algorithm" in htpasswd
is more secure than the "SHA algorithm" although the hash function
itself should be stronger.
With a rainbow table, cracking "SHA" is easily done.
A rainbow table for "MD5" + salt would be way too large to be
efficiently stored.
Hence this commit is reverted to old behaviour to avoid the clear
failure of design in SHA.
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
Signed-off-by: Arne Fitzenreiter <arne.fitzenreiter@ipfire.org>
This is a work around to prevent not working dns
resolution if the time jumps before the DNSSec signing key.
Signed-off-by: Arne Fitzenreiter <arne_f@ipfire.org>
htpasswd doesn't protect passwords very well. MD5 was used
before and now any newly created passwords will use the
SHA format.
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
unbound does not append the local domain to the request
any more (like dnsmasq did). Therefore, the client needs
to do that if desired.
Signed-off-by: Arne Fitzenreiter <arne_f@ipfire.org>
unbound has some trouble with validating DNSSEC-enabled
domains when the upstream name server is stripping signatures
from the authoritative responses.
This script now checks that, removes any broken upstream
name servers from the list and prints a warning.
If all name servers fail the test, unbound falls back
into recursor mode.
Signed-off-by: Arne Fitzenreiter <arne_f@ipfire.org>
Previously we copied the default configuration from the upstream
package and modified that. Unfortunately a patch and a sed command
changed the file which resulted in unwanted changes.
This patch removes the patch and sed command and adds a new set
of configuration files that just need to be copied to the system.
Fixes#11195
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
This commit updates krb5 to version 1.14.4
The patch is removed, because he is upstream since 1.12.2.
The samba version is incremented, to link samba against the new krb5
version. Otherwise samba for example is linked against
/usr/lib/libkdb5.so.7 but the current version is /usr/lib/libkdb5.so.8
Signed-off-by: Jonatan Schlag <jonatan.schlag@ipfire.org>
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
The virtlogd could only be restarted when the daemons run. The update.sh
script tried to restart the daemon no matter if the daemons run or not.
This behaviour produce problems.
An If statement now checks if the daemon runs or not and execute the
command that is suitable for the situation.
Fixes: #11172
Signed-off-by: Jonatan Schlag <jonatan.schlag@ipfire.org>
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
This is the update of libvirt to the latest version 2.1.
The most important change from a packager view is the new virtlogd
daemon.
This daemon handles the qemu output and wrote it to log files.
The require some changes:
- A new init script to start, stop restart the daemon called virtlogd.
The daemon is restart with SIGUSR1 (this is important because the daemon
keeps all pipelines etc. open).
This introduces a problem with the uninstall.sh install.sh script.
It is not possible to stop the daemon while virtual machines are
running, so the script update.sh execute from now not uninstall.sh and
install.sh instead it contains all steps from uninstall.sh install.sh
expect the start / stop routine for virtlogd. The daemon is just
restarted after the update, which makes sure that all changes take
effect.
- new symlinks in the uninstall.sh and install.sh script and some root
file changes because of the new virtlogd init script.
- the archive format changes from tar.gz to tar.xz
For Changelogs see:
https://libvirt.org/news-2015.htmlhttps://libvirt.org/news.html (2017 and later:
https://libvirt.org/news-2016.html )
Signed-off-by: Jonatan Schlag <jonatan.schlag@ipfire.org>
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
For details, see:
http://www.squid-cache.org/Versions/v3/3.5/changesets/
Since there were problems with "trailing white spaces" I started a new 'squid_3'
branch from scratch, based on current 'next'.
I hope this is what is needed and that it helps.
This one was built without errors and is running here without seen problems.
Best,
Matthias
Signed-off-by: Matthias Fischer <matthias.fischer@ipfire.org>
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
If the kernel module vhot_net is loaded, the performance of virtio
networking is better then without vhost_net.
So the module is loaded before libvirtd ist started to get the benefit
of vhost_net.
Signed-off-by: Jonatan Schlag <jonatan.schlag@ipfire.org>
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
