diff options
author | mhemmatp <mhemmatp@cisco.com> | 2019-10-02 01:57:27 -0700 |
---|---|---|
committer | mhemmatp <mhemmatp@cisco.com> | 2019-10-07 02:41:47 -0700 |
commit | 108c55669102931acc9bd99ca9918379722732b8 (patch) | |
tree | 4bda2e5f0d492909e711068969bbfc6bc5b41b1b /ctrl | |
parent | df63161bafa18c05318e7a87fd6422b1095e1a5a (diff) |
[HICN-289] update README
Signed-off-by: mhemmatp <mhemmatp@cisco.com>
Change-Id: I8ad3f4d636714acc10626c78c758829bc9faf273
Diffstat (limited to 'ctrl')
-rw-r--r-- | ctrl/sysrepo-plugins/README.md | 200 |
1 files changed, 56 insertions, 144 deletions
diff --git a/ctrl/sysrepo-plugins/README.md b/ctrl/sysrepo-plugins/README.md index 3f0ed2e34..0940fff67 100644 --- a/ctrl/sysrepo-plugins/README.md +++ b/ctrl/sysrepo-plugins/README.md @@ -1,6 +1,8 @@ # Sysrepo plugin for hicn-plugin (2019) -These plugins serve as a data management agent. They provide yang models via NETCONF to allow the management of hicn-light, and hicn-plugin which runs in VPP instance from out-of-box. +These plugins serve as a data management agent. They provide yang models via +NETCONF to allow the management of hicn-light, and hicn-plugin which runs in VPP +instance from out-of-box. ## Software Requirement @@ -15,7 +17,7 @@ These plugins serve as a data management agent. They provide yang models via NET ## hICN yang model You can install the yang model using the following bash script: - +``` EXIT_CODE=0 command -v sysrepoctl > /dev/null if [ $? != 0 ]; then @@ -24,53 +26,21 @@ if [ $? != 0 ]; then else sysrepoctl --install --yang=path_to_hicn_yang_model fi - -hicn.yang can be found in the yang-model. It consists of two container nodes: hicn-conf and hicn-state. One is used to hold the configuration data (i.e., hicn-conf) and one for providing the state data (i.e., hicn-state). The hicn-conf has one node, params, which contains the hICN configuration parameters. Controler can configure these parameters through the edit-config RPC call. This node can be used to enable and to initialize the hicn-plugin in VPP instance. Hicn-state container is used to provide the state data to the controler. It consists of state, strategy, strategies, route, and face-ip-params nodes with the coresponding leaves. In hicn model variety of RPCs are provided to allow controler to communicate with hicn-plugin as well as update the state data in hicn-state. Here you can find the schematic view of the described hicn model: - - -module: hicn - +--rw hicn-conf - | +--rw params - | +--rw enable_disable? boolean - | +--rw pit_max_size? int32 - | +--rw cs_max_size? int32 - | +--rw cs_reserved_app? int32 - | +--rw pit_dflt_lifetime_sec? float - | +--rw pit_max_lifetime_sec? float - | +--rw pit_min_lifetime_sec? float - +--ro hicn-state - +--ro states - | +--ro pkts_processed? uint64 - | +--ro pkts_interest_count? uint64 - | +--ro pkts_data_count? uint64 - | +--ro pkts_from_cache_count? uint64 - | +--ro pkts_no_pit_count? uint64 - | +--ro pit_expired_count? uint64 - | +--ro cs_expired_count? uint64 - | +--ro cs_lru_count? uint64 - | +--ro pkts_drop_no_buf? uint64 - | +--ro interests_aggregated? uint64 - | +--ro interests_retx? uint64 - | +--ro interests_hash_collision? uint64 - | +--ro pit_entries_count? uint64 - | +--ro cs_entries_count? uint64 - | +--ro cs_entries_ntw_count? uint64 - +--ro strategy - | +--ro description? uint8 - +--ro route - | +--ro faceids? uint16 - | +--ro strategy_id? uint32 - +--ro strategies - | +--ro n_strategies? uint8 - | +--ro strategy_id? uint32 - +--ro face-ip-params - +--ro nh_addr? uint64 - +--ro swif? uint32 - +--ro flags? uint32 - +``` +hicn.yang can be found in the yang-model. It consists of two container nodes: +hicn-conf and hicn-state. One is used to hold the configuration data (i.e., +hicn-conf) and one for providing the state data (i.e., hicn-state). The +hicn-conf has one node, params, which contains the hICN configuration +parameters. A controller can configure these parameters through the edit-config RPC +call. This node can be used to enable and to initialize the hicn-plugin in VPP +instance. Hicn-state container is used to provide the state data to the +controller. It consists of state, strategy, strategies, route, and face-ip-params +nodes with the corresponding leaves. In the hicn model a variety of RPCs are provided +to allow controller to communicate with the hicn-plugin as well as update the state +data in hicn-state. To setup the startup configuration you can use the following script: - +``` EXIT_CODE=0 command -v sysrepocfg > /dev/null if [ $? != 0 ]; then @@ -79,10 +49,10 @@ if [ $? != 0 ]; then else sysrepocfg -d startup -i path_to_startup_xml -f xml hicn fi - +``` startup.xml is placed in the yang-model. Here you can find the content: - +``` <hicn-conf xmlns="urn:sysrepo:hicn"> <params> <enable_disable>false</enable_disable> @@ -94,81 +64,17 @@ startup.xml is placed in the yang-model. Here you can find the content: <pit_min_lifetime_sec>-1</pit_min_lifetime_sec> </params> </hicn-conf> - -As can be seen, it contains the leaves of the params in hicn-conf node which is used as the startup configuration. This configuration can be changed through the controler by subscribing which changes the target to the running state. hicn yang model provides a list of RPCs which allows controler to communicate directly with the hicn-plugin. This RPCs may also cause the modification in state data. Here you can find the list of RPCs: - - rpcs: - +---x node-params-set - | +---w input - | +---w enable_disable? boolean - | +---w pit_max_size? int32 - | +---w cs_max_size? int32 - | +---w cs_reserved_app? int32 - | +---w pit_dflt_lifetime_sec? float - | +---w pit_max_lifetime_sec? float - | +---w pit_min_lifetime_sec? float - +---x node-params-get - +---x node-stat-get - +---x strategy-get - | +---w input - | +---w strategy_id? uint32 - +---x strategies-get - +---x route-get - | +---w input - | +---w prefix0? uint64 - | +---w prefix1? uint64 - | +---w len? uint8 - +---x route-del - | +---w input - | +---w prefix0? uint64 - | +---w prefix1? uint64 - | +---w len? uint8 - +---x route-nhops-add - | +---w input - | +---w prefix0? uint64 - | +---w prefix1? uint64 - | +---w len? uint8 - | +---w face_ids0? uint32 - | +---w face_ids1? uint32 - | +---w face_ids2? uint32 - | +---w face_ids3? uint32 - | +---w face_ids4? uint32 - | +---w face_ids5? uint32 - | +---w face_ids6? uint32 - | +---w n_faces? uint8 - +---x route-nhops-del - | +---w input - | +---w prefix0? uint64 - | +---w prefix1? uint64 - | +---w len? uint8 - | +---w faceid? uint16 - +---x face-ip-params-get - | +---w input - | +---w faceid? uint16 - +---x face-ip-add - | +---w input - | +---w nh_addr0? uint64 - | +---w nh_addr1? uint64 - | +---w swif? uint32 - +---x face-ip-del - | +---w input - | +---w faceid? uint16 - +---x punting-add - | +---w input - | +---w prefix0? uint64 - | +---w prefix1? uint64 - | +---w len? uint8 - | +---w swif? uint32 - +---x punting-del - +---w input - +---w prefix0? uint64 - +---w prefix1? uint64 - +---w len? uint8 - +---w swif? uint32 - - -In order to run different RPCs from controler you can use the examples in the controler_rpcs_instances.xml in the yang-model. Here you can find the content: - +``` +As can be seen, it contains the leaves of the params in hicn-conf node which is +used as the startup configuration. This configuration can be changed through the +controller by subscribing which changes the target to the running state. hicn +yang model provides a list of RPCs which allows controller to communicate +directly with the hicn-plugin. This RPCs may also cause the modification in +state data. + +In order to run different RPCs from controller you can use the examples in the +controler_rpcs_instances.xml in the yang-model. Here you can find the content: +``` <node-params-get xmlns="urn:sysrepo:hicn"/> <node-stat-get xmlns="urn:sysrepo:hicn"/> @@ -179,7 +85,6 @@ In order to run different RPCs from controler you can use the examples in the co <strategies-get xmlns="urn:sysrepo:hicn"/> - <route-get xmlns="urn:sysrepo:hicn"> <prefix0>10</prefix0> <prefix1>20</prefix1> @@ -213,7 +118,6 @@ In order to run different RPCs from controler you can use the examples in the co <faceid>40</faceid> </route-nhops-del> - <face-ip-params-get xmlns="urn:sysrepo:hicn"> <faceid>10</faceid> </face-ip-params-get> @@ -235,27 +139,33 @@ In order to run different RPCs from controler you can use the examples in the co <swif>40</swif> </punting-add> - <punting-del xmlns="urn:sysrepo:hicn"> <prefix0>10</prefix0> <prefix1>20</prefix1> <len>30</len> <swif>40</swif> </punting-del> - +``` ## Run the plugin -Firstly, verify the plugin and binary libraries are located correctly, then run the vpp through (service vpp start). Next, run the sysrepo daemon (sysrepod), for debug mode: sysrepo -d -l 4 which runs with high verbosity. Then, run the sysrepo plugin (sysrepo-plugind), for debug mode: sysrep-plugind -d -l 4 which runs with high verbosity. Now, the hicn sysrepo plugin is loaded. Then, run the netopeer2-server which serves as NETCONF server. +Firstly, verify the plugin and binary libraries are located correctly, then run +the vpp through (service vpp start). Next, run the sysrepo daemon (sysrepod), +for debug mode: sysrepo -d -l 4 which runs with high verbosity. Then, run the +sysrepo plugin (sysrepo-plugind), for debug mode: sysrep-plugind -d -l 4 which +runs with high verbosity. Now, the hicn sysrepo plugin is loaded. Then, run the +netopeer2-server which serves as NETCONF server. ## Connect from netopeer2-cli In order to connect through the netopeer client run the netopeer2-cli. Then, follow these steps: -connect --host XXX --login XXX ---> get (you can get the configuration and operational data) ---> get-config (you can get the configuratoin data) ---> edit-config --target running --config (you can modify the configuration but it needs an xml configuration input. For example, +- connect --host XXX --login XXX +- get (you can get the configuration and operational data) +- get-config (you can get the configuration data) +- edit-config --target running --config + you can modify the configuration but it needs an xml configuration input +``` <hicn-conf xmlns="urn:sysrepo:hicn"> <params> <enable_disable>false</enable_disable> @@ -267,15 +177,16 @@ connect --host XXX --login XXX <pit_min_lifetime_sec>-1</pit_min_lifetime_sec> </params> </hicn-conf> -) ---> user-rpc (you can call one of the rpc proposed by hicn model but it needs an xml input, you can pick one in controler_rpcs_instances.xml) +``` +- user-rpc (you can call one of the rpc proposed by hicn model but it needs an xml input) -## Connect from OpenDayligh (ODL) controller +## Connect from OpenDaylight (ODL) controller -In order to connect througt the OpenDaylight follow these procedure: +In order to connect through the OpenDaylight follow these procedure: - run karaf distribution (./opendayligh_installation_folder/bin/karaf) -- install the required feature list in DOL (feature:install odl-netconf-server odl-netconf-connector odl-restconf-all odl-netconf-topology or +- install the required feature list in DOL (feature:install odl-netconf-server + odl-netconf-connector odl-restconf-all odl-netconf-topology or odl-netconf-clustered-topology) - run a rest client program (e.g., postman or RESTClient) - mount the remote netopeer2-server to the OpenDaylight by the following REST API: @@ -283,7 +194,7 @@ In order to connect througt the OpenDaylight follow these procedure: PUT http://localhost:8181/restconf/config/network-topology:network-topology/topology/topology-netconf/node/hicn-node with the following body - +``` <node xmlns="urn:TBD:params:xml:ns:yang:network-topology"> <node-id>hicn-node</node-id> <host xmlns="urn:opendaylight:netconf-node-topology">Remote_NETCONF_SERVER_IP</host> @@ -293,8 +204,8 @@ with the following body <tcp-only xmlns="urn:opendaylight:netconf-node-topology">false</tcp-only> <keepalive-delay xmlns="urn:opendaylight:netconf-node-topology">1</keepalive-delay> </node> - -Note that the header files must be set to Content-Type: application/xml, Accept: application/xml. There are more options which can be set but for simplicity we keep a short configuration to mount the remote node. +``` +Note that the header files must be set to Content-Type: application/xml, Accept: application/xml. - send the operation through the following REST API: @@ -304,7 +215,9 @@ The body can be used the same as edit-config in netopeer2-cli. ## Connect from Network Services Orchestrator (NSO) -To connect NSO to the netopeer2-server, first, you need to write a NED package for your device. The procudeure to create NED for hicn is explaned in the following: +To connect NSO to the netopeer2-server, first, you need to write a NED package +for your device. The procedure to create NED for hicn is explained in the +following: Place hicn.yang model in a folder called hicn-yang-model, and follow these steps: @@ -323,7 +236,6 @@ Place hicn.yang model in a folder called hicn-yang-model, and follow these steps At this point, we are able to connect to the remote device. - ## Release note -The current version is compatible with the 19.01 VPP stable and sysrepo 0.7.7. +The current version is compatible with the 19.01 VPP stable and sysrepo 0.7.7.
\ No newline at end of file |