Bug 1931522 - Fresh UPI install on BM with bonding using OVN Kubernetes fails
Summary: Fresh UPI install on BM with bonding using OVN Kubernetes fails
Keywords:
Status: CLOSED ERRATA
Alias: None
Product: OpenShift Container Platform
Classification: Red Hat
Component: Machine Config Operator
Version: 4.6.z
Hardware: x86_64
OS: Linux
medium
high
Target Milestone: ---
: 4.8.0
Assignee: Tim Rozet
QA Contact: Michael Nguyen
URL:
Whiteboard:
Depends On:
Blocks: 1931615
TreeView+ depends on / blocked
 
Reported: 2021-02-22 15:48 UTC by Simon Belmas-Gauderic
Modified: 2024-06-14 00:25 UTC (History)
7 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
: 1931615 (view as bug list)
Environment:
Last Closed: 2021-07-27 22:47:38 UTC
Target Upstream Version:
Embargoed:


Attachments (Terms of Use)


Links
System ID Private Priority Status Summary Last Updated
Github openshift machine-config-operator pull 2424 0 None closed Bug 1931522: OVS Config: fixes detecting bond NM files with static IP 2021-04-11 15:37:18 UTC
Red Hat Product Errata RHSA-2021:2438 0 None None None 2021-07-27 22:48:13 UTC

Description Simon Belmas-Gauderic 2021-02-22 15:48:49 UTC
Description of problem:

After second boot, When MCO tries to configure network, the script configure-ovs.sh fails.


Version-Release number of selected component (if applicable):

Openshift 4.6.16

How reproducible:


Steps to Reproduce:
1. provision node as described in documentation with kernel parameters

2. Wait for image write on disk and after reboot the script configure-ovs.sh fails

Actual results:

The configure-ovs.sh script fails and prevend OVN br-ex to be created

Expected results:

The script ends successfully

Additional info:
Dracut puts `master={BOND_UUID}` in slave interfaces config files.

I tested to change uuid by interfaces names (to avoid script crash) but after restart, NetworkManager didn't set the bond up.
UUID seems to be the proper way to refer master.

The configure-ovs.sh script at line 183 select interfaces files by greping bond uuid: 
` if egrep -l --include=*.nmconnection $old_conn ${NM_CONN_PATH}/*; then`.

This matches 3 interface files (the bond and it's two slaves) because in slave `master={BOND_UUID}` is written.

The script then fails to cp 3 files in one at line 202 `cp -f ${old_conn_file} ${new_conn_file}`

Comment 13 Michael Nguyen 2021-03-03 16:16:40 UTC
Verified script changes landed in 4.8.0-0.nightly-2021-03-03-072205.


$ oc get clusterversion
NAME      VERSION                             AVAILABLE   PROGRESSING   SINCE   STATUS
version   4.8.0-0.nightly-2021-03-03-072205   True        False         17m     Cluster version is 4.8.0-0.nightly-2021-03-03-072205
$ oc get node
NAME                                         STATUS   ROLES    AGE   VERSION
ip-10-0-136-105.us-west-2.compute.internal   Ready    worker   29m   v1.20.0+2ce2be0
ip-10-0-138-238.us-west-2.compute.internal   Ready    master   38m   v1.20.0+2ce2be0
ip-10-0-173-154.us-west-2.compute.internal   Ready    master   38m   v1.20.0+2ce2be0
ip-10-0-177-34.us-west-2.compute.internal    Ready    worker   29m   v1.20.0+2ce2be0
ip-10-0-210-181.us-west-2.compute.internal   Ready    worker   30m   v1.20.0+2ce2be0
ip-10-0-220-196.us-west-2.compute.internal   Ready    master   38m   v1.20.0+2ce2be0
$ oc debug node/ip-10-0-210-181.us-west-2.compute.internal
Starting pod/ip-10-0-210-181us-west-2computeinternal-debug ...
To use host binaries, run `chroot /host`
If you don't see a command prompt, try pressing enter.
sh-4.2# chroot /host
sh-4.4# cat /usr/local/bin/configure-ovs.sh 
#!/bin/bash
set -eux
# Workaround to ensure OVS is installed due to bug in systemd Requires:
# https://bugzilla.redhat.com/show_bug.cgi?id=1888017
copy_nm_conn_files() {
  src_path="/etc/NetworkManager/system-connections-merged"
  dst_path="/etc/NetworkManager/system-connections"
  if [ -d $src_path ]; then
    echo "$src_path exists"
    fileList=$(echo {br-ex,ovs-if-br-ex,ovs-port-br-ex,ovs-if-phys0,ovs-port-phys0}.nmconnection)
    for file in ${fileList[*]}; do
      if [ ! -f $dst_path/$file ]; then
        cp $src_path/$file $dst_path/$file
      else
        echo "Skipping $file since it exists in $dst_path"
      fi
    done
  fi
}

if ! rpm -qa | grep -q openvswitch; then
  echo "Warning: Openvswitch package is not installed!"
  exit 1
fi

if [ "$1" == "OVNKubernetes" ]; then
  # Configures NICs onto OVS bridge "br-ex"
  # Configuration is either auto-detected or provided through a config file written already in Network Manager
  # key files under /etc/NetworkManager/system-connections/
  # Managing key files is outside of the scope of this script

  # if the interface is of type vmxnet3 add multicast capability for that driver
  # REMOVEME: Once BZ:1854355 is fixed, this needs to get removed.
  function configure_driver_options {
    intf=$1
    driver=$(cat "/sys/class/net/${intf}/device/uevent" | grep DRIVER | awk -F "=" '{print $2}')
    echo "Driver name is" $driver
    if [ "$driver" = "vmxnet3" ]; then
      ifconfig "$intf" allmulti
    fi
  }
  if [ -d "/etc/NetworkManager/system-connections-merged" ]; then
    NM_CONN_PATH="/etc/NetworkManager/system-connections-merged"
  else
    NM_CONN_PATH="/etc/NetworkManager/system-connections"
  fi
  iface=""
  counter=0
  # find default interface
  while [ $counter -lt 12 ]; do
    # check ipv4
    iface=$(ip route show default | awk '{ if ($4 == "dev") { print $5; exit } }')
    if [[ -n "$iface" ]]; then
      echo "IPv4 Default gateway interface found: ${iface}"
      break
    fi
    # check ipv6
    iface=$(ip -6 route show default | awk '{ if ($4 == "dev") { print $5; exit } }')
    if [[ -n "$iface" ]]; then
      echo "IPv6 Default gateway interface found: ${iface}"
      break
    fi
    counter=$((counter+1))
    echo "No default route found on attempt: ${counter}"
    sleep 5
  done

  if [ "$iface" = "br-ex" ]; then
    # handle vlans and bonds etc if they have already been
    # configured via nm key files and br-ex is already up
    ifaces=$(ovs-vsctl list-ifaces ${iface})
    for intf in $ifaces; do configure_driver_options $intf; done
    echo "Networking already configured and up for br-ex!"
    # remove bridges created by openshift-sdn
    ovs-vsctl --timeout=30 --if-exists del-br br0
    exit 0
  fi

  if [ -z "$iface" ]; then
    echo "ERROR: Unable to find default gateway interface"
    exit 1
  fi

  # find the MAC from OVS config or the default interface to use for OVS internal port
  # this prevents us from getting a different DHCP lease and dropping connection
  if ! iface_mac=$(<"/sys/class/net/${iface}/address"); then
    echo "Unable to determine default interface MAC"
    exit 1
  fi

  echo "MAC address found for iface: ${iface}: ${iface_mac}"

  # find MTU from original iface
  iface_mtu=$(ip link show "$iface" | awk '{print $5; exit}')
  if [[ -z "$iface_mtu" ]]; then
    echo "Unable to determine default interface MTU, defaulting to 1500"
    iface_mtu=1500
  else
    echo "MTU found for iface: ${iface}: ${iface_mtu}"
  fi

  # store old conn for later
  old_conn=$(nmcli --fields UUID,DEVICE conn show --active | awk "/\s${iface}\s*\$/ {print \$1}")

  extra_brex_args=""
  # check for dhcp client ids
  dhcp_client_id=$(nmcli --get-values ipv4.dhcp-client-id conn show ${old_conn})
  if [ -n "$dhcp_client_id" ]; then
    extra_brex_args+="ipv4.dhcp-client-id ${dhcp_client_id} "
  fi

  dhcp6_client_id=$(nmcli --get-values ipv6.dhcp-duid conn show ${old_conn})
  if [ -n "$dhcp6_client_id" ]; then
    extra_brex_args+="ipv6.dhcp-duid ${dhcp6_client_id} "
  fi

  # create bridge; use NM's ethernet device default route metric (100)
  if ! nmcli connection show br-ex &> /dev/null; then
    nmcli c add type ovs-bridge \
        con-name br-ex \
        conn.interface br-ex \
        802-3-ethernet.mtu ${iface_mtu} \
        802-3-ethernet.cloned-mac-address ${iface_mac} \
        ipv4.route-metric 100 \
        ipv6.route-metric 100 \
        ${extra_brex_args}
  fi

  # find default port to add to bridge
  if ! nmcli connection show ovs-port-phys0 &> /dev/null; then
    nmcli c add type ovs-port conn.interface ${iface} master br-ex con-name ovs-port-phys0
  fi

  if ! nmcli connection show ovs-port-br-ex &> /dev/null; then
    nmcli c add type ovs-port conn.interface br-ex master br-ex con-name ovs-port-br-ex
  fi

  extra_phys_args=""
  # check if this interface is a vlan, bond, or ethernet type
  if [ $(nmcli --get-values connection.type conn show ${old_conn}) == "vlan" ]; then
    iface_type=vlan
    vlan_id=$(nmcli --get-values vlan.id conn show ${old_conn})
    if [ -z "$vlan_id" ]; then
      echo "ERROR: unable to determine vlan_id for vlan connection: ${old_conn}"
      exit 1
    fi
    vlan_parent=$(nmcli --get-values vlan.parent conn show ${old_conn})
    if [ -z "$vlan_parent" ]; then
      echo "ERROR: unable to determine vlan_parent for vlan connection: ${old_conn}"
      exit 1
    fi
    extra_phys_args="dev ${vlan_parent} id ${vlan_id}"
  elif [ $(nmcli --get-values connection.type conn show ${old_conn}) == "bond" ]; then
    iface_type=bond
    # check bond options
    bond_opts=$(nmcli --get-values bond.options conn show ${old_conn})
    if [ -n "$bond_opts" ]; then
      extra_phys_args+="bond.options ${bond_opts} "
    fi
  else
    iface_type=802-3-ethernet
  fi

  # bring down any old iface
  nmcli device disconnect $iface

  if ! nmcli connection show ovs-if-phys0 &> /dev/null; then
    nmcli c add type ${iface_type} conn.interface ${iface} master ovs-port-phys0 con-name ovs-if-phys0 \
      connection.autoconnect-priority 100 802-3-ethernet.mtu ${iface_mtu} ${extra_phys_args}
  fi

  nmcli conn up ovs-if-phys0

  if ! nmcli connection show ovs-if-br-ex &> /dev/null; then
    if nmcli --fields ipv4.method,ipv6.method conn show $old_conn | grep manual; then
      echo "Static IP addressing detected on default gateway connection: ${old_conn}"
      # find and copy the old connection to get the address settings
      if egrep -l --include=*.nmconnection uuid=$old_conn ${NM_CONN_PATH}/*; then
        old_conn_file=$(egrep -l --include=*.nmconnection uuid=$old_conn ${NM_CONN_PATH}/*)
        cloned=false
      else
        echo "WARN: unable to find NM configuration file for conn: ${old_conn}. Attempting to clone conn"
        old_conn_file=${NM_CONN_PATH}/${old_conn}-clone.nmconnection
        nmcli conn clone ${old_conn} ${old_conn}-clone
        cloned=true
        if [ ! -f "$old_conn_file" ]; then
          echo "ERROR: unable to locate cloned conn file: ${old_conn_file}"
          exit 1
        fi
        echo "Successfully cloned conn to ${old_conn_file}"
      fi
      echo "old connection file found at: ${old_conn_file}"
      new_conn_file=${NM_CONN_PATH}/ovs-if-br-ex.nmconnection
      if [ -f "$new_conn_file" ]; then
        echo "WARN: existing br-ex interface file found: $new_conn_file, which is not loaded in NetworkManager...overwriting"
      fi
      cp -f ${old_conn_file} ${new_conn_file}
      restorecon ${new_conn_file}
      if $cloned; then
        nmcli conn delete ${old_conn}-clone
        rm -f ${old_conn_file}
      fi
      ovs_port_conn=$(nmcli --fields connection.uuid conn show ovs-port-br-ex | awk '{print $2}')
      br_iface_uuid=$(cat /proc/sys/kernel/random/uuid)
      # modify file to work with OVS and have unique settings
      sed -i '/^\[connection\]$/,/^\[/ s/^uuid=.*$/uuid='"$br_iface_uuid"'/' ${new_conn_file}
      sed -i '/^multi-connect=.*$/d' ${new_conn_file}
      sed -i '/^\[connection\]$/,/^\[/ s/^type=.*$/type=ovs-interface/' ${new_conn_file}
      sed -i '/^\[connection\]$/,/^\[/ s/^id=.*$/id=ovs-if-br-ex/' ${new_conn_file}
      sed -i '/^\[connection\]$/a slave-type=ovs-port' ${new_conn_file}
      sed -i '/^\[connection\]$/a master='"$ovs_port_conn" ${new_conn_file}
      if grep 'interface-name=' ${new_conn_file} &> /dev/null; then
        sed -i '/^\[connection\]$/,/^\[/ s/^interface-name=.*$/interface-name=br-ex/' ${new_conn_file}
      else
        sed -i '/^\[connection\]$/a interface-name=br-ex' ${new_conn_file}
      fi
      if ! grep 'cloned-mac-address=' ${new_conn_file} &> /dev/null; then
        sed -i '/^\[ethernet\]$/,/^\[/ s/^cloned-mac-address=.*$/cloned-mac-address='"$iface_mac"'/' ${new_conn_file}
      else
        sed -i '/^\[ethernet\]$/a cloned-mac-address='"$iface_mac" ${new_conn_file}
      fi
      if grep 'mtu=' ${new_conn_file} &> /dev/null; then
        sed -i '/^\[ethernet\]$/,/^\[/ s/^mtu=.*$/mtu='"$iface_mtu"'/' ${new_conn_file}
      else
        sed -i '/^\[ethernet\]$/a mtu='"$iface_mtu" ${new_conn_file}
      fi
      cat <<EOF >> ${new_conn_file}
[ovs-interface]
type=internal
EOF
      nmcli c load ${new_conn_file}
      echo "Loaded new ovs-if-br-ex connection file: ${new_conn_file}"
    else
      nmcli c add type ovs-interface slave-type ovs-port conn.interface br-ex master ovs-port-br-ex con-name \
        ovs-if-br-ex 802-3-ethernet.mtu ${iface_mtu} 802-3-ethernet.cloned-mac-address ${iface_mac} \
        ipv4.route-metric 100 ipv6.route-metric 100
    fi
  fi

  # wait for DHCP to finish, verify connection is up
  counter=0
  while [ $counter -lt 5 ]; do
    sleep 5
    # check if connection is active
    if nmcli --fields GENERAL.STATE conn show ovs-if-br-ex | grep -i "activated"; then
      echo "OVS successfully configured"
      copy_nm_conn_files
      ip a show br-ex
      ip route show
      configure_driver_options ${iface}
      exit 0
    fi
    counter=$((counter+1))
  done

  echo "WARN: OVS did not succesfully activate NM connection. Attempting to bring up connections"
  counter=0
  while [ $counter -lt 5 ]; do
    if nmcli conn up ovs-if-br-ex; then
      echo "OVS successfully configured"
      copy_nm_conn_files
      ip a show br-ex
      ip route show
      configure_driver_options ${iface}
      exit 0
    fi
    sleep 5
    counter=$((counter+1))
  done

  echo "ERROR: Failed to activate ovs-if-br-ex NM connection"
  # if we made it here networking isnt coming up, revert for debugging
  set +e
  nmcli conn down ovs-if-br-ex
  nmcli conn down ovs-if-phys0
  nmcli conn up $old_conn
  exit 1
elif [ "$1" == "OpenShiftSDN" ]; then
  # Revert changes made by /usr/local/bin/configure-ovs.sh.
  # Remove OVS bridge "br-ex". Use the default NIC for cluster network.
  iface=""
  if nmcli connection show ovs-port-phys0 &> /dev/null; then
    iface=$(nmcli --get-values connection.interface-name connection show ovs-port-phys0)
    nmcli c del ovs-port-phys0 
  fi

  if nmcli connection show ovs-if-phys0 &> /dev/null; then
    nmcli c del ovs-if-phys0
  fi

  if nmcli connection show ovs-port-br-ex &> /dev/null; then
    nmcli c del ovs-port-br-ex
  fi

  if nmcli connection show ovs-if-br-ex &> /dev/null; then
    nmcli c del ovs-if-br-ex
  fi

  if nmcli connection show br-ex &> /dev/null; then
    nmcli c del br-ex
  fi

  rm -f /etc/NetworkManager/system-connections/{br-ex,ovs-if-br-ex,ovs-port-br-ex,ovs-if-phys0,ovs-port-phys0}.nmconnection
  # remove bridges created by ovn-kubernetes, try to delete br-ex again in case NM fail to talk to ovsdb
  ovs-vsctl --timeout=30 --if-exists del-br br-int -- --if-exists del-br br-local -- --if-exists del-br br-ex

  if [[ -n "$iface" ]]; then
    nmcli device connect $iface
  fi
fi
sh-4.4#

Comment 18 Lucas López Montero 2021-04-27 08:17:01 UTC
KCS article created to describe this problem: https://access.redhat.com/solutions/5995781.

Comment 22 errata-xmlrpc 2021-07-27 22:47:38 UTC
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory (Moderate: OpenShift Container Platform 4.8.2 bug fix and security update), and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHSA-2021:2438

Comment 23 Red Hat Bugzilla 2023-09-15 01:01:51 UTC
The needinfo request[s] on this closed bug have been removed as they have been unresolved for 500 days


Note You need to log in before you can comment on or make changes to this bug.