From 34908bb59ac6999030bf313e31d873ac09dc3faa Mon Sep 17 00:00:00 2001 From: Mark Gurevich Date: Wed, 22 Feb 2017 16:29:34 -0500 Subject: [PATCH 1/5] update bonding configuration instructions --- .../common/deployment/cfg_network.rst | 20 ++++++++++++++++--- 1 file changed, 17 insertions(+), 3 deletions(-) diff --git a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network.rst b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network.rst index aa0e78f03..5c8e1e42b 100644 --- a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network.rst +++ b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network.rst @@ -19,6 +19,8 @@ The following example set the xCAT properties for compute node ``cn1`` to achiev Define attributes in the ``nics`` table --------------------------------------- +Chose one of three methods described below: + #. Using the ``mkdef`` or ``chdef`` commands a. Compute node ``cn1`` has two physical NICs: ``eth2`` and ``eth3`` :: @@ -30,7 +32,7 @@ Define attributes in the ``nics`` table chdef cn1 nictypes.bond0=bond \ nicdevices.bond0="eth2|eth3" - c. Fom ``bond0``, create 2 VLANs: ``bond0.1`` and ``bond0.2`` :: + c. From ``bond0``, create 2 VLANs: ``bond0.1`` and ``bond0.2`` :: chdef cn1 nictypes.bond0.1=vlan \ nictypes.bond0.2=vlan \ @@ -102,6 +104,11 @@ Use the ``chdef`` command to add/modify the networks in the ``networks`` table : chdef -t network net10 net=10.0.0.0 mask=255.0.0.0 mgtifname=eth0 chdef -t network net20 net=20.0.0.0 mask=255.0.0.0 mgtifname=eth1 +For bonding add :: + + chdef -t network net40 net=20.0.0.0 mask=255.0.0.0 mgtifname=eth1 + chdef cn1 nicnetworks.bond0=net40 + Add ``confignetwork`` into the node's postscripts list ------------------------------------------------------ @@ -110,12 +117,19 @@ Using below command to add ``confignetwork`` into the node's postscripts list :: chdef cn1 -p postscripts=confignetwork -During OS deployment on compute node, ``confignetwork`` will be run in postscript. -If the compute node has OS, use ``updatenode`` command to run ``confignetwork`` :: +During OS deployment on compute node, ``confignetwork`` postscript will be executed. +If the compute node is already running, use ``updatenode`` command to run ``confignetwork`` postscript without rebooting the node:: updatenode cn1 -P confignetwork +Verify bonding mode +------------------- +Login to compute node and check bonding options in ``/etc/sysconfig/network-scripts/ifcfg-bond0`` file :: + BONDING_OPTS="mode=802.3ad xmit_hash_policy=layer2+3" +The ``mode=802.3ad`` requires additional configuration on the switch. ``mode=2`` can be used for bonding without additional switch configuration. If changes are made to ``/etc/sysconfig/network-scripts/ifcfg-bond0`` file, restart network service :: + + systemctl restart network.service From 5006290b623d1984b993da1dea5dd60e6f85177f Mon Sep 17 00:00:00 2001 From: Mark Gurevich Date: Thu, 23 Feb 2017 14:25:45 -0500 Subject: [PATCH 2/5] Refactor adapter bonding instructions --- .../common/deployment/cfg_network.rst | 135 +----------------- .../common/deployment/cfg_network_bond.rst | 55 +++++++ .../common/deployment/cfg_network_vlan.rst | 118 +++++++++++++++ .../common/deployment/cfg_second_adapter.rst | 9 +- 4 files changed, 182 insertions(+), 135 deletions(-) create mode 100644 docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_bond.rst create mode 100644 docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_vlan.rst diff --git a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network.rst b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network.rst index 5c8e1e42b..0c3d779c2 100644 --- a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network.rst +++ b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network.rst @@ -1,135 +1,8 @@ Advanced Networking Configuration ================================= -The ``confignetwork`` postscript can be used to configure the network interfaces on the compute nodes to support VLAN, BONDs, and BRIDGES. In order to use the ``confignetwork`` postscript, the following attributes must be configured for the node in the ``nics`` table: +.. toctree:: + :maxdepth: 2 - * ``nicips`` - * ``nictypes`` - * ``nicnetworks`` - * ``nicdevices`` - resolves the relationship among the physical network intereface devices - -The following example set the xCAT properties for compute node ``cn1`` to achieve the following network configuration using the ``confignetwork`` postscript: - - * Compute node ``cn1`` has two physical NICs: eth2 and eth3 - * Bond eth2 and eth3 as ``bond0`` - * From ``bond0``, create 2 VLANs: ``bond0.1`` and ``bond0.2`` - * Make bridge ``br1`` using ``bond0.1`` with IP (10.0.0.1) - * Make bridge ``br2`` using ``bond0.2`` with IP (20.0.0.1) - -Define attributes in the ``nics`` table ---------------------------------------- - -Chose one of three methods described below: - -#. Using the ``mkdef`` or ``chdef`` commands - - a. Compute node ``cn1`` has two physical NICs: ``eth2`` and ``eth3`` :: - - chdef cn1 nictypes.eth2=ethernet nictypes.eth3=ethernet - - b. Define ``bond0`` and bond ``eth2`` and ``eth3`` as ``bond0`` :: - - chdef cn1 nictypes.bond0=bond \ - nicdevices.bond0="eth2|eth3" - - c. From ``bond0``, create 2 VLANs: ``bond0.1`` and ``bond0.2`` :: - - chdef cn1 nictypes.bond0.1=vlan \ - nictypes.bond0.2=vlan \ - nicdevices.bond0.1=bond0 \ - nicdevices.bond0.2=bond0 - - d. Create bridge ``br1`` using ``bond0.1`` with IP (10.0.0.1) :: - - chdef cn1 nictypes.br1=bridge \ - nicdevices.br1=bond0.1 \ - nicips.br1=10.0.0.1 \ - nicnetworks.br1="net10" - - e. Create bridge ``br2`` using ``bond0.2`` with IP (20.0.0.1) :: - - chdef cn1 nictypes.br2=bridge \ - nicdevices.br2=bond0.2 \ - nicips.br2=20.0.0.1 \ - nicnetworks.br2="net20" - -#. Using an xCAT stanza file - - - Prepare a stanza file ``.stanza`` with content similiar to the following: :: - - # - cn1: - objtype=node - arch=x86_64 - groups=kvm,vm,all - nicdevices.br1=bond0.1 - nicdevices.br2=bond0.2 - nicdevices.bond0.1=bond0 - nicdevices.bond0.2=bond0 - nicdevices.bond0=eth2|eth3 - nictypes.eth2=ethernet - nictypes.eth3=ethernet - nictypes.bond0=bond - nictypes.bond0.1=vlan - nictypes.bond0.2=vlan - nictypes.br1=bridge - nictypes.br2=bridge - nicips.br1=10.0.0.1 - nicips.br2=20.0.0.1 - nicnetworks.br1=net10 - nicnetworks.br2=net20 - - - Using the ``mkdef -z`` option, define the stanza file to xCAT: :: - - cat .stanza | mkdef -z - -#. Using ``tabedit`` to edit the ``nics`` database table directly - - The ``tabedit`` command opens the specified xCAT database table in a ``vi`` like editor and allows the user to edit any text and write the changes back to the database table. - - After changing the content of the ``nics`` table, here is the result from ``tabdump nics`` :: - - # tabdump nics - #node,nicips,nichostnamesuffixes,nichostnameprefixes,nictypes,niccustomscripts,nicnetworks,nicaliases,nicextraparams,nicdevices,comments,disable - "cn1","br1!10.0.0.1,br2!20.0.0.1",,,"br1!bridge,eth2!ethernet,eth3!ethernet,bond0.2!vlan,bond0!bond,br2!bridge,bond0.1!vlan",,"br1!net10,br2!net20",,,"br1!bond0.1,bond0!eth2|eth3,bond0.2!bond0,bond0.1!bond0,br2!bond0.2",, - -Add network object into the networks table ------------------------------------------- - -The ``nicnetworks`` attribute only defines the nic that uses the IP address. -Other information about the network should be defined in the ``networks`` table. - -Use the ``chdef`` command to add/modify the networks in the ``networks`` table :: - - chdef -t network net10 net=10.0.0.0 mask=255.0.0.0 mgtifname=eth0 - chdef -t network net20 net=20.0.0.0 mask=255.0.0.0 mgtifname=eth1 - -For bonding add :: - - chdef -t network net40 net=20.0.0.0 mask=255.0.0.0 mgtifname=eth1 - chdef cn1 nicnetworks.bond0=net40 - -Add ``confignetwork`` into the node's postscripts list ------------------------------------------------------- - -Using below command to add ``confignetwork`` into the node's postscripts list :: - - chdef cn1 -p postscripts=confignetwork - - -During OS deployment on compute node, ``confignetwork`` postscript will be executed. -If the compute node is already running, use ``updatenode`` command to run ``confignetwork`` postscript without rebooting the node:: - - updatenode cn1 -P confignetwork - - -Verify bonding mode -------------------- - -Login to compute node and check bonding options in ``/etc/sysconfig/network-scripts/ifcfg-bond0`` file :: - - BONDING_OPTS="mode=802.3ad xmit_hash_policy=layer2+3" - -The ``mode=802.3ad`` requires additional configuration on the switch. ``mode=2`` can be used for bonding without additional switch configuration. If changes are made to ``/etc/sysconfig/network-scripts/ifcfg-bond0`` file, restart network service :: - - systemctl restart network.service +.. include:: ../../../common/deployment/cfg_network_vlan.rst +.. include:: ../../../common/deployment/cfg_network_bond.rst diff --git a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_bond.rst b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_bond.rst new file mode 100644 index 000000000..ad3bdd89c --- /dev/null +++ b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_bond.rst @@ -0,0 +1,55 @@ +Configure Two Bonded Adapters +----------------------------- + +The following example set the xCAT properties for compute node ``cn1`` to create: + + * Compute node ``cn1`` has two physical NICs: eth2 and eth3 + * Bond eth2 and eth3 as ``bond0`` + +Define attributes in the ``nics`` table +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + + +#. Using the ``mkdef`` or ``chdef`` commands + + a. Compute node ``cn1`` has two physical NICs: ``eth2`` and ``eth3`` :: + + chdef cn1 nictypes.eth2=ethernet nictypes.eth3=ethernet + + b. Define ``bond0`` and bond ``eth2`` and ``eth3`` as ``bond0`` :: + + chdef cn1 nictypes.bond0=bond \ + nicdevices.bond0="eth2|eth3" + +Add network object into the networks table +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Use the ``chdef`` command to add/modify the networks in the ``networks`` table :: + + chdef -t network net40 net=20.0.0.0 mask=255.0.0.0 mgtifname=eth1 + chdef cn1 nicnetworks.bond0=net40 + +Add ``confignetwork`` into the node's postscripts list +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Use command below to add ``confignetwork`` into the node's postscripts list :: + + chdef cn1 -p postscripts=confignetwork + + +During OS deployment on compute node, ``confignetwork`` postscript will be executed. +If the compute node is already running, use ``updatenode`` command to run ``confignetwork`` postscript without rebooting the node:: + + updatenode cn1 -P confignetwork + + +Verify bonding mode +~~~~~~~~~~~~~~~~~~~ + +Login to compute node cn1 and check bonding options in ``/etc/sysconfig/network-scripts/ifcfg-bond0`` file :: + + BONDING_OPTS="mode=802.3ad xmit_hash_policy=layer2+3" + +The ``mode=802.3ad`` requires additional configuration on the switch. ``mode=2`` can be used for bonding without additional switch configuration. If changes are made to ``/etc/sysconfig/network-scripts/ifcfg-bond0`` file, restart network service :: + + systemctl restart network.service diff --git a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_vlan.rst b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_vlan.rst new file mode 100644 index 000000000..a4d637634 --- /dev/null +++ b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_vlan.rst @@ -0,0 +1,118 @@ +Configure BOND, VLAN and BRIDGES +-------------------------------- + +The ``confignetwork`` postscript can be used to configure the network interfaces on the compute nodes to support VLAN, BONDs, and BRIDGES. In order to use the ``confignetwork`` postscript, the following attributes must be configured for the node in the ``nics`` table: + + * ``nicips`` + * ``nictypes`` + * ``nicnetworks`` + * ``nicdevices`` - resolves the relationship among the physical network intereface devices + +The following example set the xCAT properties for compute node ``cn1`` to achieve the following network configuration using the ``confignetwork`` postscript: + + * Compute node ``cn1`` has two physical NICs: eth2 and eth3 + * Bond eth2 and eth3 as ``bond0`` + * From ``bond0``, create 2 VLANs: ``bond0.1`` and ``bond0.2`` + * Make bridge ``br1`` using ``bond0.1`` with IP (10.0.0.1) + * Make bridge ``br2`` using ``bond0.2`` with IP (20.0.0.1) + +Define attributes in the ``nics`` table +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Chose one of three methods described below: + +#. Using the ``mkdef`` or ``chdef`` commands + + a. Compute node ``cn1`` has two physical NICs: ``eth2`` and ``eth3`` :: + + chdef cn1 nictypes.eth2=ethernet nictypes.eth3=ethernet + + b. Define ``bond0`` and bond ``eth2`` and ``eth3`` as ``bond0`` :: + + chdef cn1 nictypes.bond0=bond \ + nicdevices.bond0="eth2|eth3" + + c. From ``bond0``, create 2 VLANs: ``bond0.1`` and ``bond0.2`` :: + + chdef cn1 nictypes.bond0.1=vlan \ + nictypes.bond0.2=vlan \ + nicdevices.bond0.1=bond0 \ + nicdevices.bond0.2=bond0 + + d. Create bridge ``br1`` using ``bond0.1`` with IP (10.0.0.1) :: + + chdef cn1 nictypes.br1=bridge \ + nicdevices.br1=bond0.1 \ + nicips.br1=10.0.0.1 \ + nicnetworks.br1="net10" + + e. Create bridge ``br2`` using ``bond0.2`` with IP (20.0.0.1) :: + + chdef cn1 nictypes.br2=bridge \ + nicdevices.br2=bond0.2 \ + nicips.br2=20.0.0.1 \ + nicnetworks.br2="net20" + +#. Using an xCAT stanza file + + - Prepare a stanza file ``.stanza`` with content similiar to the following: :: + + # + cn1: + objtype=node + arch=x86_64 + groups=kvm,vm,all + nicdevices.br1=bond0.1 + nicdevices.br2=bond0.2 + nicdevices.bond0.1=bond0 + nicdevices.bond0.2=bond0 + nicdevices.bond0=eth2|eth3 + nictypes.eth2=ethernet + nictypes.eth3=ethernet + nictypes.bond0=bond + nictypes.bond0.1=vlan + nictypes.bond0.2=vlan + nictypes.br1=bridge + nictypes.br2=bridge + nicips.br1=10.0.0.1 + nicips.br2=20.0.0.1 + nicnetworks.br1=net10 + nicnetworks.br2=net20 + + - Using the ``mkdef -z`` option, define the stanza file to xCAT: :: + + cat .stanza | mkdef -z + +#. Using ``tabedit`` to edit the ``nics`` database table directly + + The ``tabedit`` command opens the specified xCAT database table in a ``vi`` like editor and allows the user to edit any text and write the changes back to the database table. + + After changing the content of the ``nics`` table, here is the result from ``tabdump nics`` :: + + # tabdump nics + #node,nicips,nichostnamesuffixes,nichostnameprefixes,nictypes,niccustomscripts,nicnetworks,nicaliases,nicextraparams,nicdevices,comments,disable + "cn1","br1!10.0.0.1,br2!20.0.0.1",,,"br1!bridge,eth2!ethernet,eth3!ethernet,bond0.2!vlan,bond0!bond,br2!bridge,bond0.1!vlan",,"br1!net10,br2!net20",,,"br1!bond0.1,bond0!eth2|eth3,bond0.2!bond0,bond0.1!bond0,br2!bond0.2",, + +Add network object into the networks table +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +The ``nicnetworks`` attribute only defines the nic that uses the IP address. +Other information about the network should be defined in the ``networks`` table. + +Use the ``chdef`` command to add/modify the networks in the ``networks`` table :: + + chdef -t network net10 net=10.0.0.0 mask=255.0.0.0 mgtifname=eth0 + chdef -t network net20 net=20.0.0.0 mask=255.0.0.0 mgtifname=eth1 + +Add ``confignetwork`` into the node's postscripts list +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Using below command to add ``confignetwork`` into the node's postscripts list :: + + chdef cn1 -p postscripts=confignetwork + +During OS deployment on compute node, ``confignetwork`` postscript will be executed. +If the compute node is already running, use ``updatenode`` command to run ``confignetwork`` postscript without rebooting the node:: + + updatenode cn1 -P confignetwork + diff --git a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_second_adapter.rst b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_second_adapter.rst index a6abdd3e9..76475cf02 100644 --- a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_second_adapter.rst +++ b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_second_adapter.rst @@ -47,7 +47,7 @@ There are 3 ways to complete this operation. The ``tabedit`` command opens the specified xCAT database table in a vi like editor and allows the user to edit any text and write the changes back to the database table. - *WARNING* Using the ``tabedit`` command is not the recommended method because it is tedious and error prone. + *WARNING:* Using the ``tabedit`` command is not the recommended method because it is tedious and error prone. After changing the content of the ``nics`` table, here is the result from ``tabdump nics`` :: @@ -70,7 +70,7 @@ After you have defined the configuration information in any of the ways above, r Add confignics into the node's postscripts list ----------------------------------------------- -Using below command to add confignics into the node's postscripts list :: +Use command below to add confignics into the node's postscripts list :: chdef cn1 -p postscripts=confignics @@ -78,7 +78,7 @@ By default, confignics does not configure the install nic. if need, using flag " chdef cn1 -p prostscripts="confignics -s" -Option "-s" write the install nic's information into configuration file for persistance. All install nic's data defined in nics table will be written also. +Option "-s" writes the install nic's information into configuration file for persistance. All install nic's data defined in nics table will be written also. Add network object into the networks table @@ -103,6 +103,7 @@ Option -r to remove the undefined NICS If the compute node's nics were configured by ``confignics`` and the nics configuration changed in the nics table, user the ``confignics -r`` to remove the undefined nic. For example, if on a compute node the ``eth0``, ``eth1``, and ``eth2`` nics were configured: :: + # ifconfig eth0 Link encap:Ethernet HWaddr 00:14:5e:d9:6c:e6 ... @@ -114,7 +115,7 @@ For example, if on a compute node the ``eth0``, ``eth1``, and ``eth2`` nics were Delete the eth2 definition in nics table using the ``chdef`` command. Then run the following to remove the undefined ``eth2`` nic on the compute node: :: - # updatenode -P "confignics -r" + updatenode -P "confignics -r" The result should have ``eth2`` disabled: :: From 16d259dae52ab28249acfd992f778b1101bf549a Mon Sep 17 00:00:00 2001 From: Mark Gurevich Date: Fri, 24 Feb 2017 08:34:27 -0500 Subject: [PATCH 3/5] Added missing chdef statement --- .../manage_clusters/common/deployment/cfg_network_bond.rst | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_bond.rst b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_bond.rst index ad3bdd89c..d7360b127 100644 --- a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_bond.rst +++ b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_bond.rst @@ -5,6 +5,7 @@ The following example set the xCAT properties for compute node ``cn1`` to create * Compute node ``cn1`` has two physical NICs: eth2 and eth3 * Bond eth2 and eth3 as ``bond0`` + * Assign ip ``40.0.0.1`` to the bonded interface ``bond0`` Define attributes in the ``nics`` table ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ @@ -18,15 +19,15 @@ Define attributes in the ``nics`` table b. Define ``bond0`` and bond ``eth2`` and ``eth3`` as ``bond0`` :: - chdef cn1 nictypes.bond0=bond \ - nicdevices.bond0="eth2|eth3" + chdef cn1 nictypes.bond0=bond nicdevices.bond0="eth2|eth3" + chdef cn1 nicips.bond0=40.0.0.1 Add network object into the networks table ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Use the ``chdef`` command to add/modify the networks in the ``networks`` table :: - chdef -t network net40 net=20.0.0.0 mask=255.0.0.0 mgtifname=eth1 + chdef -t network net40 net=40.0.0.0 mask=255.0.0.0 mgtifname=eth1 chdef cn1 nicnetworks.bond0=net40 Add ``confignetwork`` into the node's postscripts list From c26a67e61e519d209222af9e81a681dab301fe4a Mon Sep 17 00:00:00 2001 From: Mark Gurevich Date: Wed, 1 Mar 2017 15:58:10 -0500 Subject: [PATCH 4/5] Update after review --- .../common/deployment/cfg_network.rst | 4 ++-- .../common/deployment/cfg_network_bond.rst | 6 ++++-- .../common/deployment/cfg_network_vlan.rst | 17 +++++++++-------- .../common/deployment/cfg_second_adapter.rst | 4 ++-- 4 files changed, 17 insertions(+), 14 deletions(-) diff --git a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network.rst b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network.rst index 0c3d779c2..c36c0e3c2 100644 --- a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network.rst +++ b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network.rst @@ -1,5 +1,5 @@ -Advanced Networking Configuration -================================= +Advanced Networking Configuration - confignetwork +================================================= .. toctree:: :maxdepth: 2 diff --git a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_bond.rst b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_bond.rst index d7360b127..5bf46b9db 100644 --- a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_bond.rst +++ b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_bond.rst @@ -51,6 +51,8 @@ Login to compute node cn1 and check bonding options in ``/etc/sysconfig/network- BONDING_OPTS="mode=802.3ad xmit_hash_policy=layer2+3" -The ``mode=802.3ad`` requires additional configuration on the switch. ``mode=2`` can be used for bonding without additional switch configuration. If changes are made to ``/etc/sysconfig/network-scripts/ifcfg-bond0`` file, restart network service :: - systemctl restart network.service +* ``mode=802.3ad`` requires additional configuration on the switch. +* ``mode=2`` can be used for bonding without additional switch configuration. + +If changes are made to ``/etc/sysconfig/network-scripts/ifcfg-bond0`` file, restart network service. diff --git a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_vlan.rst b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_vlan.rst index a4d637634..a245ff172 100644 --- a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_vlan.rst +++ b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_vlan.rst @@ -1,17 +1,17 @@ Configure BOND, VLAN and BRIDGES -------------------------------- -The ``confignetwork`` postscript can be used to configure the network interfaces on the compute nodes to support VLAN, BONDs, and BRIDGES. In order to use the ``confignetwork`` postscript, the following attributes must be configured for the node in the ``nics`` table: +The ``confignetwork`` postscript can be used to configure the network interfaces on the compute nodes to support VLAN, BONDs, and BRIDGES. In order for the ``confignetwork`` postscript to run successfully, the following attributes must be configured for the node in the ``nics`` table: * ``nicips`` * ``nictypes`` * ``nicnetworks`` - * ``nicdevices`` - resolves the relationship among the physical network intereface devices + * ``nicdevices`` - resolves the relationship among the physical network interface devices The following example set the xCAT properties for compute node ``cn1`` to achieve the following network configuration using the ``confignetwork`` postscript: - * Compute node ``cn1`` has two physical NICs: eth2 and eth3 - * Bond eth2 and eth3 as ``bond0`` + * Compute node ``cn1`` has two physical NICs: ``eth2`` and ``eth3`` + * Bond ``eth2`` and ``eth3`` as ``bond0`` * From ``bond0``, create 2 VLANs: ``bond0.1`` and ``bond0.2`` * Make bridge ``br1`` using ``bond0.1`` with IP (10.0.0.1) * Make bridge ``br2`` using ``bond0.2`` with IP (20.0.0.1) @@ -93,8 +93,10 @@ Chose one of three methods described below: #node,nicips,nichostnamesuffixes,nichostnameprefixes,nictypes,niccustomscripts,nicnetworks,nicaliases,nicextraparams,nicdevices,comments,disable "cn1","br1!10.0.0.1,br2!20.0.0.1",,,"br1!bridge,eth2!ethernet,eth3!ethernet,bond0.2!vlan,bond0!bond,br2!bridge,bond0.1!vlan",,"br1!net10,br2!net20",,,"br1!bond0.1,bond0!eth2|eth3,bond0.2!bond0,bond0.1!bond0,br2!bond0.2",, -Add network object into the networks table -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +Define the additional networks to xCAT +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +If this is a new network being created on the compute nodes, an entry needs to be created into the xCAT database. The ``nicnetworks`` attribute only defines the nic that uses the IP address. Other information about the network should be defined in the ``networks`` table. @@ -107,11 +109,10 @@ Use the ``chdef`` command to add/modify the networks in the ``networks`` table : Add ``confignetwork`` into the node's postscripts list ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -Using below command to add ``confignetwork`` into the node's postscripts list :: +Use the following command to add ``confignetwork`` into postscript list to execute on reboot: :: chdef cn1 -p postscripts=confignetwork -During OS deployment on compute node, ``confignetwork`` postscript will be executed. If the compute node is already running, use ``updatenode`` command to run ``confignetwork`` postscript without rebooting the node:: updatenode cn1 -P confignetwork diff --git a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_second_adapter.rst b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_second_adapter.rst index 76475cf02..492361fc5 100644 --- a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_second_adapter.rst +++ b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_second_adapter.rst @@ -1,5 +1,5 @@ -Configure Additional Network Interfaces -======================================= +Configure Additional Network Interfaces - confignics +==================================================== The **nics** table and the **confignics** postscript can be used to automatically configure additional network interfaces (mutltiple ethernets adapters, InfiniBand, etc) on the nodes as they are being deployed. From 88323b74009e1b1dfd33732f97f6a76bd973b13d Mon Sep 17 00:00:00 2001 From: Mark Gurevich Date: Thu, 2 Mar 2017 14:05:11 -0500 Subject: [PATCH 5/5] Final review changes --- .../common/deployment/cfg_network.rst | 2 ++ .../common/deployment/cfg_network_bond.rst | 2 +- .../common/deployment/cfg_network_vlan.rst | 16 +++------------- 3 files changed, 6 insertions(+), 14 deletions(-) diff --git a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network.rst b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network.rst index c36c0e3c2..e11d21d67 100644 --- a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network.rst +++ b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network.rst @@ -1,6 +1,8 @@ Advanced Networking Configuration - confignetwork ================================================= +**Note:** ``confignetwork`` postscript is only supported on RHEL releases. + .. toctree:: :maxdepth: 2 diff --git a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_bond.rst b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_bond.rst index 5bf46b9db..b4c31a981 100644 --- a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_bond.rst +++ b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_bond.rst @@ -27,7 +27,7 @@ Add network object into the networks table Use the ``chdef`` command to add/modify the networks in the ``networks`` table :: - chdef -t network net40 net=40.0.0.0 mask=255.0.0.0 mgtifname=eth1 + chdef -t network net40 net=40.0.0.0 mask=255.0.0.0 chdef cn1 nicnetworks.bond0=net40 Add ``confignetwork`` into the node's postscripts list diff --git a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_vlan.rst b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_vlan.rst index a245ff172..3ddd9606f 100644 --- a/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_vlan.rst +++ b/docs/source/guides/admin-guides/manage_clusters/common/deployment/cfg_network_vlan.rst @@ -19,7 +19,7 @@ The following example set the xCAT properties for compute node ``cn1`` to achiev Define attributes in the ``nics`` table ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -Chose one of three methods described below: +Chose one of two methods described below: #. Using the ``mkdef`` or ``chdef`` commands @@ -83,16 +83,6 @@ Chose one of three methods described below: cat .stanza | mkdef -z -#. Using ``tabedit`` to edit the ``nics`` database table directly - - The ``tabedit`` command opens the specified xCAT database table in a ``vi`` like editor and allows the user to edit any text and write the changes back to the database table. - - After changing the content of the ``nics`` table, here is the result from ``tabdump nics`` :: - - # tabdump nics - #node,nicips,nichostnamesuffixes,nichostnameprefixes,nictypes,niccustomscripts,nicnetworks,nicaliases,nicextraparams,nicdevices,comments,disable - "cn1","br1!10.0.0.1,br2!20.0.0.1",,,"br1!bridge,eth2!ethernet,eth3!ethernet,bond0.2!vlan,bond0!bond,br2!bridge,bond0.1!vlan",,"br1!net10,br2!net20",,,"br1!bond0.1,bond0!eth2|eth3,bond0.2!bond0,bond0.1!bond0,br2!bond0.2",, - Define the additional networks to xCAT ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ @@ -103,8 +93,8 @@ Other information about the network should be defined in the ``networks`` table. Use the ``chdef`` command to add/modify the networks in the ``networks`` table :: - chdef -t network net10 net=10.0.0.0 mask=255.0.0.0 mgtifname=eth0 - chdef -t network net20 net=20.0.0.0 mask=255.0.0.0 mgtifname=eth1 + chdef -t network net10 net=10.0.0.0 mask=255.0.0.0 + chdef -t network net20 net=20.0.0.0 mask=255.0.0.0 Add ``confignetwork`` into the node's postscripts list ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~