cephadm: Added command to list RDMA devices

Added 'cephadm list-rdma' command to list RDMA devices and their netdev interfaces

Fixes: https://tracker.ceph.com/issues/75189
Signed-off-by: Shweta Bhosale <Shweta.Bhosale1@ibm.com>

diff --git a/doc/man/8/cephadm.rst b/doc/man/8/cephadm.rst
index 0a42850bd59b29a3cbf44ad18bced4bdb54e6446..ca43ca06dda74f77d650d71202a0d739a37c9609 100644 (file)
--- a/doc/man/8/cephadm.rst
+++ b/doc/man/8/cephadm.rst
@@ -13,7 +13,7 @@ Synopsis
 |               [--log-dir LOG_DIR] [--logrotate-dir LOGROTATE_DIR]
 |               [--unit-dir UNIT_DIR] [--verbose] [--timeout TIMEOUT]
 |               [--retry RETRY] [--no-container-init]
-|               {version,pull,inspect-image,ls,list-networks,adopt,rm-daemon,rm-cluster,run,shell,enter,ceph-volume,unit,logs,bootstrap,deploy,check-host,prepare-host,add-repo,rm-repo,install,list-images,update-osd-service}
+|               {version,pull,inspect-image,ls,list-networks,list-rdma,adopt,rm-daemon,rm-cluster,run,shell,enter,ceph-volume,unit,logs,bootstrap,deploy,check-host,prepare-host,add-repo,rm-repo,install,list-images,update-osd-service}
 |               ...
 
 
@@ -25,6 +25,8 @@ Synopsis
 
 | **cephadm** **list-networks**
 
+| **cephadm** **list-rdma**
+
 | **cephadm** **adopt** [-h] --name NAME --style STYLE [--cluster CLUSTER]
 |                       [--legacy-dir LEGACY_DIR] [--config-json CONFIG_JSON]
 |                       [--skip-firewalld] [--skip-pull]
@@ -347,6 +349,10 @@ list-networks
 
 list IP networks
 
+list-rdma
+---------
+
+list RDMA devices and their netdev interfaces
 
 ls
 --

diff --git a/src/cephadm/cephadm.py b/src/cephadm/cephadm.py
index 7038d4214c8c73108c737e6222079e68cdec9f1b..2539ea46999c6832ec38804c4f18bebe665503df 100755 (executable)
--- a/src/cephadm/cephadm.py
+++ b/src/cephadm/cephadm.py
@@ -155,7 +155,7 @@ from cephadmlib.decorators import (
     executes_early,
     require_image
 )
-from cephadmlib.host_facts import HostFacts, list_networks
+from cephadmlib.host_facts import HostFacts, list_networks, list_rdma
 from cephadmlib.ssh import authorize_ssh_key, check_ssh_connectivity
 from cephadmlib.daemon_form import (
     DaemonForm,
@@ -3563,6 +3563,11 @@ def command_list_networks(ctx):
 
     print(json.dumps(r, indent=4, default=serialize_sets))
 
+
+def command_list_rdma(ctx: CephadmContext) -> None:
+    r = list_rdma(ctx)
+    print(json.dumps(r, indent=4))
+
 ##################################
 
 
@@ -5011,6 +5016,10 @@ def _get_parser():
         'list-networks', help='list IP networks')
     parser_list_networks.set_defaults(func=command_list_networks)
 
+    parser_list_rdma = subparsers.add_parser(
+        'list-rdma', help='list RDMA devices and their netdev interfaces')
+    parser_list_rdma.set_defaults(func=command_list_rdma)
+
     parser_adopt = subparsers.add_parser(
         'adopt', help='adopt daemon deployed with a different tool')
     parser_adopt.set_defaults(func=command_adopt)

diff --git a/src/cephadm/cephadmlib/host_facts.py b/src/cephadm/cephadmlib/host_facts.py
index af5d2e23a67bcfc340046363a41910f6100bbe09..9d7098f8be61c47088ddf5cc51236a31aaa8e84c 100644 (file)
--- a/src/cephadm/cephadmlib/host_facts.py
+++ b/src/cephadm/cephadmlib/host_facts.py
@@ -860,6 +860,49 @@ def list_networks(ctx):
     return res
 
 
+def list_rdma(ctx: CephadmContext) -> List[Dict[str, str]]:
+    """List RDMA devices by parsing 'rdma link show' output.
+    Returns a list of dicts with keys: link, state, physical_state, netdev.
+    Returns empty list if rdma tool is not installed or command fails.
+    """
+    execstr: Optional[str] = find_executable('rdma')
+    if not execstr:
+        logger.error("'rdma' command not found, no RDMA devices listed")
+        return []
+    try:
+        out, _, _ = call_throws(
+            ctx,
+            [execstr, 'link', 'show'],
+            verbosity=CallVerbosity.QUIET_UNLESS_ERROR,
+        )
+    except Exception as e:
+        logger.error('rdma link show failed: %s', e)
+        return []
+    # Format: link <name> state <state> physical_state <phys> netdev <netdev>
+    pattern = re.compile(
+        r'link\s+(\S+)\s+state\s+(\S+)\s+physical_state\s+(\S+)\s+netdev\s+'
+        r'(\S+)'
+    )
+    result: List[Dict[str, str]] = []
+    for line in out.splitlines():
+        line = line.strip()
+        if not line:
+            continue
+        m = pattern.search(line)
+        if m:
+            result.append(
+                {
+                    'link': m.group(1),
+                    'state': m.group(2),
+                    'physical_state': m.group(3),
+                    'netdev': m.group(4),
+                }
+            )
+        else:
+            logger.debug("Skipped RDMA device '%s', as pattern did not match", line)
+    return result
+
+
 def _list_ipv4_networks(
     ctx: CephadmContext,
 ) -> Dict[str, Dict[str, Set[str]]]: