[FRR] bgpcfgd Wait for mgmtd initial config load (#24655)

Why I did it
Work item tracking
Microsoft ADO (number only):
34848755
How I did it
Wait for mgmtd initial config load to avoid "Lock already taken on DS" error

How to verify it
2025 Dec 2 00:56:56.302357 bjw-can-7060-1 NOTICE bgp#bgpcfgd: Checking mgmtd datastore readiness...
2025 Dec 2 00:56:56.390985 bjw-can-7060-1 NOTICE bgp#bgpcfgd: mgmtd datastores are ready (attempt 1)

Signed-off-by: Feng Pan <fenpan@microsoft.com>

Author: lipxu

src/sonic-bgpcfgd/bgpcfgd/main.py

@@ -1,8 +1,10 @@
 import os
 import signal
+import subprocess
 import sys
 import syslog
 import threading
+import time
 import traceback
 
 from swsscommon.swsscommon import ConfigDBConnector
@@ -11,7 +13,7 @@
 
 from .config import ConfigMgr
 from .directory import Directory
-from .log import log_notice, log_crit
+from .log import log_notice, log_crit, log_warn
 from .managers_advertise_rt import AdvertiseRouteMgr
 from .managers_aggregate_address import AggregateAddressMgr, BGP_AGGREGATE_ADDRESS_TABLE_NAME
 from .managers_allow_list import BGPAllowListMgr
@@ -43,6 +45,23 @@ def do_work():
     thr.start()
     frr = FRR(["bgpd", "zebra", "staticd"])
     frr.wait_for_daemons(seconds=20)
+
+    # Wait for mgmtd initial config load to avoid "Lock already taken on DS" error
+    log_notice("Checking mgmtd datastore readiness...")
+    for attempt in range(10):  # Max ~5 seconds
+        try:
+            out = subprocess.check_output(['vtysh', '-c', 'show mgmt datastore all'],
+                                        stderr=subprocess.DEVNULL, timeout=2, text=True)
+            # Check if any datastore is locked (FRR topotest approach)
+            locked_lines = [line for line in out.splitlines() if 'Locked:' in line and 'True' in line]
+            if not locked_lines:
+                log_notice("mgmtd datastores are ready (attempt %d)" % (attempt + 1))
+                break
+            else:
+                log_warn("mgmtd datastores still locked (attempt %d): %s" % (attempt + 1, ', '.join(locked_lines)))
+        except Exception as e:
+            log_warn("mgmtd check failed (attempt %d): %s" % (attempt + 1, str(e)))
+        time.sleep(0.5)
     #
     common_objs = {
         'directory': Directory(),

src/sonic-frr/patch/0097-mgmt-Note-that-a-DS-is-locked-or-not-in-output.patch

@@ -0,0 +1,42 @@
+From ac505e5f4528f6ce85673e518d1df0c7c10a6ac1 Mon Sep 17 00:00:00 2001
+From: Donald Sharp <sharpd@nvidia.com>
+Date: Fri, 11 Jul 2025 19:19:15 -0400
+Subject: [PATCH] mgmt: Note that a DS is locked or not in output
+
+Modify the `show mgmt datastore ..` command to note
+if the data store is locked and by which session-id.
+
+Signed-off-by: Donald Sharp <sharpd@nvidia.com>
+---
+ mgmtd/mgmt_ds.c | 6 ++++++
+ 1 file changed, 6 insertions(+)
+
+diff --git a/mgmtd/mgmt_ds.c b/mgmtd/mgmt_ds.c
+index 91592df22f..a0e6c83393 100644
+--- a/mgmtd/mgmt_ds.c
++++ b/mgmtd/mgmt_ds.c
+@@ -514,15 +514,21 @@ void mgmt_ds_dump_tree(struct vty *vty, struct mgmt_ds_ctx *ds_ctx,
+ 
+ void mgmt_ds_status_write_one(struct vty *vty, struct mgmt_ds_ctx *ds_ctx)
+ {
++	uint64_t session_id;
++	bool locked;
++
+ 	if (!ds_ctx) {
+ 		vty_out(vty, "    >>>>> Datastore Not Initialized!\n");
+ 		return;
+ 	}
+ 
++	locked = mgmt_ds_is_locked(ds_ctx, &session_id);
+ 	vty_out(vty, "  DS: %s\n", mgmt_ds_id2name(ds_ctx->ds_id));
+ 	vty_out(vty, "    DS-Hndl: \t\t\t%p\n", ds_ctx);
+ 	vty_out(vty, "    Config: \t\t\t%s\n",
+ 		ds_ctx->config_ds ? "True" : "False");
++	vty_out(vty, "    Locked: \t\t\t%s Session-ID: %Lu\n", locked ? "True" : "False",
++		locked ? session_id : 0);
+ }
+ 
+ void mgmt_ds_status_write(struct vty *vty)
+-- 
+2.34.1
+

src/sonic-frr/patch/series

@@ -56,3 +56,4 @@
 0094-zebra-fix-yang-data-for-mcast-group.patch
 0095-bgpd-Crash-due-to-usage-of-freed-up-evpn_overlay-att.patch
 0096-bgpd-Notify-all-incoming-outgoing-on-peer-group-noti.patch
+0097-mgmt-Note-that-a-DS-is-locked-or-not-in-output.patch