diff --git a/server/api/api/endpoints/client_spec.py b/server/api/api/endpoints/client_spec.py
index 18ec04763b2..4ae7fcc46b4 100644
--- a/server/api/api/endpoints/client_spec.py
+++ b/server/api/api/endpoints/client_spec.py
@@ -19,11 +19,12 @@
 import mlrun.common.schemas
 import server.api.api.utils
 import server.api.crud
+import server.api.utils.helpers
 
 router = APIRouter()
 
 
-@server.api.api.utils.lru_cache_with_ttl(maxsize=32, ttl_seconds=60 * 5)
+@server.api.utils.helpers.lru_cache_with_ttl(maxsize=32, ttl_seconds=60 * 5)
 def get_cached_client_spec(
     client_version: typing.Optional[str] = Header(
         None, alias=mlrun.common.schemas.HeaderNames.client_version
diff --git a/server/api/api/utils.py b/server/api/api/utils.py
index 4acec73d5a0..b7be33a9a89 100644
--- a/server/api/api/utils.py
+++ b/server/api/api/utils.py
@@ -15,10 +15,8 @@
 import asyncio
 import collections
 import copy
-import functools
 import json
 import re
-import time
 import traceback
 import typing
 import uuid
@@ -67,40 +65,6 @@ def log_and_raise(status=HTTPStatus.BAD_REQUEST.value, **kw):
     raise HTTPException(status_code=status, detail=kw)
 
 
-def lru_cache_with_ttl(maxsize=128, typed=False, ttl_seconds=60):
-    """
-    Thread-safety least-recently used cache with time-to-live (ttl_seconds) limit.
-    https://stackoverflow.com/a/71634221/5257501
-    """
-
-    class Result:
-        __slots__ = ("value", "death")
-
-        def __init__(self, value, death):
-            self.value = value
-            self.death = death
-
-    def decorator(func):
-        @functools.lru_cache(maxsize=maxsize, typed=typed)
-        def cached_func(*args, **kwargs):
-            value = func(*args, **kwargs)
-            death = time.monotonic() + ttl_seconds
-            return Result(value, death)
-
-        @functools.wraps(func)
-        def wrapper(*args, **kwargs):
-            result = cached_func(*args, **kwargs)
-            if result.death < time.monotonic():
-                result.value = func(*args, **kwargs)
-                result.death = time.monotonic() + ttl_seconds
-            return result.value
-
-        wrapper.cache_clear = cached_func.cache_clear
-        return wrapper
-
-    return decorator
-
-
 def log_path(project, uid) -> Path:
     return project_logs_path(project) / uid
 
diff --git a/server/api/utils/clients/iguazio.py b/server/api/utils/clients/iguazio.py
index 954167cb258..4a37c1bde88 100644
--- a/server/api/utils/clients/iguazio.py
+++ b/server/api/utils/clients/iguazio.py
@@ -35,6 +35,7 @@
 import mlrun.errors
 import mlrun.utils.helpers
 import mlrun.utils.singleton
+import server.api.utils.helpers
 import server.api.utils.projects.remotes.leader as project_leader
 from mlrun.utils import get_in, logger
 
@@ -340,6 +341,7 @@ def is_sync(self):
         """
         return True
 
+    @server.api.utils.helpers.lru_cache_with_ttl(maxsize=1, ttl_seconds=60 * 2)
     def try_get_grafana_service_url(self, session: str) -> typing.Optional[str]:
         """
         Try to find a ready grafana app service, and return its URL
diff --git a/server/api/utils/helpers.py b/server/api/utils/helpers.py
index c2cf8091931..93a17f16479 100644
--- a/server/api/utils/helpers.py
+++ b/server/api/utils/helpers.py
@@ -14,7 +14,9 @@
 #
 import asyncio
 import datetime
+import functools
 import re
+import time
 from typing import Optional
 
 import semver
@@ -136,3 +138,37 @@ def string_to_timedelta(date_str, raise_on_error=True):
         return None
 
     return datetime.timedelta(seconds=seconds)
+
+
+def lru_cache_with_ttl(maxsize=128, typed=False, ttl_seconds=60):
+    """
+    Thread-safety least-recently used cache with time-to-live (ttl_seconds) limit.
+    https://stackoverflow.com/a/71634221/5257501
+    """
+
+    class Result:
+        __slots__ = ("value", "death")
+
+        def __init__(self, value, death):
+            self.value = value
+            self.death = death
+
+    def decorator(func):
+        @functools.lru_cache(maxsize=maxsize, typed=typed)
+        def cached_func(*args, **kwargs):
+            value = func(*args, **kwargs)
+            death = time.monotonic() + ttl_seconds
+            return Result(value, death)
+
+        @functools.wraps(func)
+        def wrapper(*args, **kwargs):
+            result = cached_func(*args, **kwargs)
+            if result.death < time.monotonic():
+                result.value = func(*args, **kwargs)
+                result.death = time.monotonic() + ttl_seconds
+            return result.value
+
+        wrapper.cache_clear = cached_func.cache_clear
+        return wrapper
+
+    return decorator
diff --git a/tests/api/utils/clients/test_iguazio.py b/tests/api/utils/clients/test_iguazio.py
index 98e8efd94cb..dd15f59e511 100644
--- a/tests/api/utils/clients/test_iguazio.py
+++ b/tests/api/utils/clients/test_iguazio.py
@@ -215,6 +215,40 @@ async def test_get_grafana_service_url_success(
     assert grafana_url == expected_grafana_url
 
 
+@pytest.mark.parametrize("iguazio_client", ("async", "sync"), indirect=True)
+@pytest.mark.asyncio
+async def test_get_grafana_service_url_cache(
+    api_url: str,
+    iguazio_client: server.api.utils.clients.iguazio.Client,
+    requests_mock: requests_mock_package.Mocker,
+):
+    expected_grafana_url = (
+        "https://grafana.default-tenant.app.hedingber-301-1.iguazio-cd2.com"
+    )
+    grafana_service = {
+        "spec": {"kind": "grafana"},
+        "status": {
+            "state": "ready",
+            "urls": [
+                {"kind": "http", "url": "https-has-precedence"},
+                {"kind": "https", "url": expected_grafana_url},
+            ],
+        },
+    }
+    response_body = _generate_app_services_manifests_body([grafana_service])
+    requests_mock.get(f"{api_url}/api/app_services_manifests", json=response_body)
+    grafana_url = await maybe_coroutine(
+        iguazio_client.try_get_grafana_service_url("session-cookie")
+    )
+    assert grafana_url == expected_grafana_url
+
+    grafana_url = await maybe_coroutine(
+        iguazio_client.try_get_grafana_service_url("session-cookie")
+    )
+    assert requests_mock.called_once
+    assert grafana_url == expected_grafana_url
+
+
 @pytest.mark.parametrize("iguazio_client", ("async", "sync"), indirect=True)
 @pytest.mark.asyncio
 async def test_get_grafana_service_url_ignoring_disabled_service(