Clevis stubbornly keeps trying the first Tang Server (even when not Ready) for unlocking even though other Servers are Online and available

[luckylinux]

I originally reported the Issue on Ubuntu Launchpad BUG Tracker since that is where I first noticed it.

However, seems to occur on every Platform.

Essentially I have 4 x Tang Servers that register their Key in one LUKS Keyslot respectively (besides the Manual Passphrase/Password).

The issue is, if the First Tang Server (Tang1) is NOT ready (in the specific case it is "Booting" but NOT unlocked - for security reasons the Tang Servers have Full Disk Encryption for / and must be unlocked Manually), then Clevis will keep trying Tang1, despite Tang2/Tang3/Tang4 being available, online and unlocked.

Tang1 is "stuck" with "Please enter the passphrase for xxxxxx" boot Message (typical of LUKS without any Tang/Clevis/other automated unlocking). I just checked and, in that phase, the Tang1 Host Networking is DOWN (Tang1 doesn't even reply to ping / ICMP Packets).

Eventually, Clevis will contact one of the other Servers (after 5 Minutes / 300 Seconds), but shouldn't this be done much faster ?

I would expect Clevis to be smart enough to query Tang2/Tang3/Tang4 or even better, to query Tang Servers in a round-Robin manner ("random"). Basically "Tang 1 Decryption Fails, move on to Tang2". Or, as stated before, do so in a round-Robin/random Query manner.

[sarroutbi]

Hello @luckylinux . Can you please post what kind of configuration are you using? Are you binding the device multiple times or are you using sss pin? Depending on the case, as a quick fix, you might use the other alternative to check if that works as expected

[luckylinux]

Hi @sarroutbi.

Not sure what "configuration" I am using to be Perfectly honest. I followed up a Tang+Clevis tutorial around a Year ago and had been using it like this ever since.

This is a Script I use for my root on LUKS installs, that will install Keys for ALL configured Tang Servers:

#!/bin/bash

# Determine toolpath if not set already
relativepath="../" # Define relative path to go from this script to the root level of the tool
if [[ ! -v toolpath ]]; then scriptpath=$(cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd ); toolpath=$(realpath --canonicalize-missing $scriptpath/$relativepath); fi

# Load configuration
source $toolpath/config.sh

# Setup CLEVIS for automated disk unlocking
add_rfc3442_hook() {
  cat << EOF > /etc/initramfs-tools/hooks/add-rfc3442-dhclient-hook
#!/bin/sh

PREREQ=""

prereqs()
{
        echo "\$PREREQ"
}

case \$1 in
prereqs)
        prereqs
        exit 0
        ;;
esac

if [ ! -x /sbin/dhclient ]; then
        exit 0
fi

. /usr/share/initramfs-tools/scripts/functions
. /usr/share/initramfs-tools/hook-functions

# Source: https://github.com/latchset/clevis/issues/148#issuecomment-882103016
#    and: https://github.com/latchset/clevis/issues/148#issuecomment-882103016
local_libdir="/lib/x86_64-linux-gnu"
local_found="" lib="" f=""
for lib in libnss_files libnss_dns libresolv; do
    local_found=""
    for f in "\${local_libdir}/\${lib}.so".?; do
        [ -e "\${f}" ] || continue
        [ "\${verbose}" = "y" ] && echo "dns: \${lib}: \${f}"
        copy_file library "\${f}"
        local_found="\${f}"
    done
    [ -n "\${local_found}" ] || echo "WARNING: no \${local_libdir}/\${lib}.? file" 1>&2
done

mkdir -p \$DESTDIR/etc/dhcp/dhclient-exit-hooks.d/
cp -a /etc/dhcp/dhclient-exit-hooks.d/rfc3442-classless-routes \$DESTDIR/etc/dhcp/dhclient-exit-hooks.d/
EOF

  chmod +x /etc/initramfs-tools/hooks/add-rfc3442-dhclient-hook
}

# Install hook
add_rfc3442_hook

# Update APT Lists
apt-get update

# Install clevis on the system and add clevis to the initramfs
apt-get install --yes clevis clevis-luks clevis-initramfs cryptsetup-initramfs

# Ask for password
read -s -p "Enter encryption password: " password

# For each keyserver
counter=1
for keyserver in "${keyservers[@]}"
do
     # Get TANG Server Key
     curl -sfg http://$keyserver/adv -o /tmp/keyserver-$counter.jws

     # Check which keys are currently used via CLEVIS
     list_device1=$(clevis luks list -d $device1-part${root_num})
     list_device2=$(clevis luks list -d $device2-part${root_num})

     # Bind device to the TANG server via CLEVIS
     # Device 1
     if [[ "${list_device1}" == *"${keyserver}"* ]]
     then
         echo "Keyserver <$keyserver> is already installed"
     else
         echo "Install Keyserver <$keyserver> onto $device1 LUKS Header"
         echo $password | clevis luks bind -d $device1-part${root_num} tang "{\"url\": \"http://$keyserver\" , \"adv\": \"/tmp/keyserver-$counter.jws\" }"
     fi

     # Device 2
     if [[ "${list_device2}" == *"${keyserver}"* ]]
     then
         echo "Keyserver <$keyserver> is already installed"
     else
          echo "Install Keyserver <$keyserver> onto $device2 LUKS Header"
          echo $password | clevis luks bind -d $device2-part${root_num} tang "{\"url\": \"http://$keyserver\" , \"adv\": \"/tmp/keyserver-$counter.jws\" }"
     fi

     # Increment counter
     counter=$((counter+1))
done

# Clear password from memory
unset $password

# Update initramfs
update-initramfs -c -k all

# Get information
cryptsetup luksDump $device1-part${root_num}
cryptsetup luksDump $device2-part${root_num}
clevis luks list -d $device1-part${root_num}
clevis luks list -d $device2-part${root_num}