CORTX-29979: Multiple Data pod deployment: Support Node group in CDF (#…

…2116) 1. Multiple Data pod deployment: Support Node group in CDF 2. Hare builds are failing for main and custom-ci branches (#2122) Solution: 1: Added support for 'node_group' in CDF format 2: The issue was with the version of charset-normalizer, a package that aiohttp (a package required by HARE) was pulling in as a depend. We have now set the version of charset-normalizer to 2.0.12 - which works with aiohttp 3.8.1 as required by HARE. Signed-off-by: Swapnil Gaonkar <swapnil.gaonkar@seagate.com> Signed-off-by: pavankrishnat <pavan.k.thunuguntla@seagate.com> Signed-off-by: Deepak Nayak <deepak.nayak@seagate.com>
Seagate · Jun 22, 2022 · 0824c85 · 0824c85
1 parent 9267e6a
commit 0824c85
Show file tree

Hide file tree

Showing 17 changed files with 121 additions and 24 deletions.
diff --git a/Makefile b/Makefile
@@ -12,6 +12,8 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 #
+#
+#
 # For any questions about this software or licensing,
 # please email opensource@seagate.com or cortx-questions@seagate.com.
 

diff --git a/cfgen/cfgen b/cfgen/cfgen
@@ -179,6 +179,7 @@ class ShowSchema(argparse.Action):
 create_aux: <bool> # optional, defaults to false if not present in CDF
 nodes:
   - hostname: <str>    # [user@]hostname; e.g., localhost, pod-c1
+    node_group: <str>  # name of the node_group in which this node is present
     data_iface: <str>  # name of network interface; e.g., eth1, eth1:c1
     data_iface_type: tcp|o2ib  # type of network interface;
                                # optional, defaults to "tcp"

diff --git a/cfgen/dhall/types/NodeDesc.dhall b/cfgen/dhall/types/NodeDesc.dhall
@@ -19,6 +19,7 @@
 -}
 
 { hostname : Text
+, node_group: Optional Text 
 , machine_id : Optional Text
 , processorcount: Optional Natural
 , memorysize_mb: Optional Double

diff --git a/cfgen/examples/ci-boot1-2ios.yaml b/cfgen/examples/ci-boot1-2ios.yaml
@@ -1,5 +1,6 @@
 nodes:
   - hostname: localhost
+    node_group: localhost
     data_iface: eth1
     data_iface_type: tcp
     transport_type: libfab

diff --git a/cfgen/examples/ci-boot2-1confd.yaml b/cfgen/examples/ci-boot2-1confd.yaml
@@ -1,5 +1,6 @@
 nodes:
   - hostname: ssu1
+    node_group: ssu1
     data_iface: eth1
     transport_type: libfab
     m0_servers:
@@ -18,6 +19,7 @@ nodes:
       - name: m0_client_other  # name of the motr client
         instances: 2   # Number of instances, this host will run
   - hostname: ssu2
+    node_group: ssu2
     data_iface: eth1
     transport_type: libfab
     m0_servers:

diff --git a/cfgen/examples/ci-boot2.yaml b/cfgen/examples/ci-boot2.yaml
@@ -1,5 +1,6 @@
 nodes:
   - hostname: ssu1
+    node_group: ssu1
     data_iface: eth1
     transport_type: libfab
     m0_servers:
@@ -18,6 +19,7 @@ nodes:
       - name: m0_client_other  # name of the motr client
         instances: 2   # Number of instances, this host will run
   - hostname: ssu2
+    node_group: ssu2
     data_iface: eth1
     transport_type: libfab
     m0_servers:

diff --git a/cfgen/examples/ci-boot3.yaml b/cfgen/examples/ci-boot3.yaml
@@ -1,5 +1,6 @@
 nodes:
   - hostname: ssu1
+    node_group: ssu1
     data_iface: eth1
     transport_type: libfab
     m0_servers:
@@ -18,6 +19,7 @@ nodes:
       - name: m0_client_other  # name of the motr client
         instances: 2   # Number of instances, this host will run
   - hostname: ssu2
+    node_group: ssu2
     data_iface: eth1
     transport_type: libfab
     m0_servers:
@@ -36,6 +38,7 @@ nodes:
       - name: m0_client_other  # name of the motr client
         instances: 2   # Number of instances, this host will run
   - hostname: ssu3
+    node_group: ssu3
     data_iface: eth1
     transport_type: libfab
     m0_servers:

diff --git a/cfgen/examples/ldr1-cluster.yaml b/cfgen/examples/ldr1-cluster.yaml
@@ -3,6 +3,7 @@
 
 nodes:
   - hostname: pod-c1        # [user@]hostname
+    node_group: pod-c1
     data_iface: eth1_c1     # name of data network interface
     data_iface_type: o2ib   # LNet type of network interface (optional);
                             # supported values: "tcp" (default), "o2ib"
@@ -21,6 +22,7 @@ nodes:
       - name: m0_client_other  # name of the motr client
         instances: 2   # Number of instances, this host will run
   - hostname: pod-c2
+    node_group: pod-c2
     data_iface: eth1_c2
     data_iface_type: o2ib
     transport_type: libfab

diff --git a/cfgen/examples/multipools.yaml b/cfgen/examples/multipools.yaml
@@ -1,5 +1,6 @@
 nodes:
   - hostname: srvnode-1
+    node_group: srvnode-1
     data_iface: enp175s0f1_c1
     data_iface_type: o2ib
     transport_type: libfab
@@ -33,6 +34,7 @@ nodes:
     #     port: 21500
     #   m0_client_s3: 22500
   - hostname: srvnode-2
+    node_group: srvnode-2
     data_iface: enp175s0f1_c2
     data_iface_type: o2ib
     transport_type: libfab

diff --git a/cfgen/examples/singlenode.yaml b/cfgen/examples/singlenode.yaml
@@ -3,6 +3,7 @@
 
 nodes:
   - hostname: localhost     # [user@]hostname
+    node_group: localhost
     data_iface: eth1        # name of data network interface
     data_iface_ip_addr: null
     transport_type: libfab

diff --git a/cfgen/tests/singlenode.dhall b/cfgen/tests/singlenode.dhall
@@ -26,6 +26,7 @@ in
 , nodes =
     [ { hostname = "localhost"
       , machine_id = None Text
+      , node_group = Some "localhost"
       , memorysize_mb = None Double
       , processorcount = None Natural
       , transport_type = "libfab"

diff --git a/provisioning/miniprov/hare_mp/cdf.py b/provisioning/miniprov/hare_mp/cdf.py
@@ -128,8 +128,14 @@ def _create_node_descriptions(self) -> List[NodeDesc]:
             if machine not in machines:
                 machines.append(machine)
 
+        is_node_group_supported: bool = False
+        local_node_group = self.utils.get_node_group(conf.get_machine_id(),
+                                                     allow_null=True)
+
+        if local_node_group:
+            is_node_group_supported = True
         for machine in machines:
-            nodes.append(self._create_node(machine))
+            nodes.append(self._create_node(machine, is_node_group_supported))
         return nodes
 
     # cluster>storage_set[N]>durability>{type}>data/parity/spare
@@ -495,14 +501,17 @@ def _get_node_clients(self, machine_id: str) -> Iterator[M0ClientDesc]:
                         name=Text(name),
                         instances=no_instances)
 
-    def _create_node(self, machine_id: str) -> NodeDesc:
+    def _create_node(self, machine_id: str,
+                     is_node_group_supported: bool) -> NodeDesc:
         store = self.provider
 
         hostname = self.utils.get_hostname(machine_id)
-        # node>{machine-id}>name
+        node_group = None
         iface = self._get_iface(machine_id)
         servers = None
         if(self.utils.is_motr_io_present(machine_id)):
+            if is_node_group_supported:
+                node_group = self.utils.get_node_group(machine_id)
             # Currently, there is 1 m0d per cvg.
             # We will create 1 IO service entry in CDF per cvg.
             # An IO service entry will use data  and metadat devices
@@ -543,6 +552,7 @@ def _create_node(self, machine_id: str) -> NodeDesc:
 
         return NodeDesc(
             hostname=Text(hostname),
+            node_group=Maybe(Text(str(node_group)), 'Text'),
             machine_id=Maybe(Text(machine_id), 'Text'),
             processorcount=Maybe(node_facts['processorcount'], 'Natural'),
             memorysize_mb=Maybe(node_facts['memorysize_mb'], 'Double'),

diff --git a/provisioning/miniprov/hare_mp/dhall/gencdf.dhall b/provisioning/miniprov/hare_mp/dhall/gencdf.dhall
@@ -38,6 +38,7 @@ let ServerPort =
 
 let NodeInfo =
       { hostname : Text
+      , node_group : Optional Text
       , machine_id : Optional Text
       , processorcount : Optional Natural
       , memorysize_mb : Optional Double
@@ -85,6 +86,7 @@ let toNodeDesc
     : NodeInfo -> T.NodeDesc
     =     \(n : NodeInfo)
       ->  { hostname = n.hostname
+          , node_group = n.node_group
           , machine_id = n.machine_id
           , processorcount = n.processorcount
           , memorysize_mb = n.memorysize_mb

diff --git a/provisioning/miniprov/hare_mp/templates/hare.config.conf.tmpl.3-node.sample b/provisioning/miniprov/hare_mp/templates/hare.config.conf.tmpl.3-node.sample
@@ -38,6 +38,7 @@
       "cluster_id": "my-cluster",
       "hostname": "ssc-vm-1623.colo.seagate.com",
       "name": "srvnode-1",
+      "node_group": "ssc-vm-1623.colo.seagate.com",
       "type": "storage_node",
       "components": [
         { "name": "hare" },
@@ -86,6 +87,7 @@
       "cluster_id": "my-cluster",
       "hostname": "ssc-vm-1624.colo.seagate.com",
       "name": "srvnode-2",
+      "node_group": "ssc-vm-1624.colo.seagate.com",
       "type": "storage_node",
       "network": {
         "data": {
@@ -129,6 +131,7 @@
       "cluster_id": "my-cluster",
       "hostname": "ssc-vm-1625.colo.seagate.com",
       "name": "srvnode-3",
+      "node_group": "ssc-vm-1625.colo.seagate.com",
       "type": "storage_node",
       "network": {
         "data": {

diff --git a/provisioning/miniprov/hare_mp/types.py b/provisioning/miniprov/hare_mp/types.py
@@ -133,6 +133,7 @@ class NetworkPorts(DhallTuple):
 @dataclass(repr=False)
 class NodeDesc(DhallTuple):
     hostname: Text
+    node_group: Maybe[Text]
     machine_id: Maybe[Text]
     processorcount: Maybe[int]
     memorysize_mb: Maybe[int]

diff --git a/provisioning/miniprov/hare_mp/utils.py b/provisioning/miniprov/hare_mp/utils.py
@@ -349,6 +349,18 @@ def save_ssl_config(self):
         })
         self.kv.kv_put('ssl/hax', ssl_hax)
 
+    # Provisioner will be generating node_group from each data pod and place
+    # gconf copy into consul. Hence from consul for all data pods node_group
+    # value will be available.
+    @func_log(func_enter, func_leave)
+    @repeat_if_fails()
+    def get_node_group(self, machine_id: str, allow_null: bool = False):
+        key = f'conf/node>{machine_id}>node_group'
+        node_group = self.kv.kv_get(key, allow_null=allow_null)
+        if node_group:
+            return node_group['Value'].decode()
+        return None
+
 
 class LogWriter:
     def __init__(self, logger: logging.Logger, logging_handler):