cephadm: adjust the ingress ha proxy health check interval

Currently health checker uses default value of 2s, it is send list
bucket request for every 2s. This seems to be frequent and need to
adjust properly. Hence introducing new setting health_check_interval in
the ingress spec for haproxy.

Signed-off-by: Jiffin Tony Thottan <thottanjiffin@gmail.com>
Apply suggestions from code review

Co-authored-by: Adam King <47704447+adk3798@users.noreply.github.com>
Signed-off-by: Jiffin Tony Thottan  <thottanjiffin@gmail.com>

diff --git a/doc/cephadm/services/rgw.rst b/doc/cephadm/services/rgw.rst
index 20ec39a88dd1a8fa84e5d13d9c31ca716017d957..ed0b149365a55943e9c2a355c89ae7eee300234a 100644 (file)
--- a/doc/cephadm/services/rgw.rst
+++ b/doc/cephadm/services/rgw.rst
@@ -246,6 +246,7 @@ It is a yaml format file with the following properties:
       virtual_interface_networks: [ ... ]       # optional: list of CIDR networks
       use_keepalived_multicast: <bool>          # optional: Default is False.
       vrrp_interface_network: <string>/<string> # optional: ex: 192.168.20.0/24
+      health_check_interval: <string>           # optional: Default is 2s.
       ssl_cert: |                               # optional: SSL certificate and key
         -----BEGIN CERTIFICATE-----
         ...
@@ -273,6 +274,7 @@ It is a yaml format file with the following properties:
       monitor_port: <integer>             # ex: 1967, used by haproxy for load balancer status
       virtual_interface_networks: [ ... ] # optional: list of CIDR networks
       first_virtual_router_id: <integer>  # optional: default 50
+      health_check_interval: <string>     # optional: Default is 2s.
       ssl_cert: |                         # optional: SSL certificate and key
         -----BEGIN CERTIFICATE-----
         ...
@@ -321,6 +323,9 @@ where the properties of this service specification are:
     keepalived will have different virtual_router_id. In the case of using ``virtual_ips_list``,
     each IP will create its own virtual router. So the first one will have ``first_virtual_router_id``,
     second one will have ``first_virtual_router_id`` + 1, etc. Valid values go from 1 to 255.
+* ``health_check_interval``
+    Default is 2 seconds. This parameter can be used to set the interval between health checks
+    for the haproxy with the backend servers.
 
 .. _ingress-virtual-ip:
 

diff --git a/src/pybind/mgr/cephadm/services/ingress.py b/src/pybind/mgr/cephadm/services/ingress.py
index 55be3045466abe3c9d130b2cabc8c871bec2fb73..5edd2517dc01b0d74360301de87ec307999c07de 100644 (file)
--- a/src/pybind/mgr/cephadm/services/ingress.py
+++ b/src/pybind/mgr/cephadm/services/ingress.py
@@ -187,6 +187,7 @@ class IngressService(CephService):
                 'monitor_port': daemon_spec.ports[1] if daemon_spec.ports else spec.monitor_port,
                 'local_host_ip': host_ip,
                 'default_server_opts': server_opts,
+                'health_check_interval': spec.health_check_interval or '2s',
             }
         )
         config_files = {

diff --git a/src/pybind/mgr/cephadm/templates/services/ingress/haproxy.cfg.j2 b/src/pybind/mgr/cephadm/templates/services/ingress/haproxy.cfg.j2
index c114a8cba11527c99fca7bcf10349fcb3aeb2e42..9a0309ab409d5f786b9d19f66275b6986efac4d9 100644 (file)
--- a/src/pybind/mgr/cephadm/templates/services/ingress/haproxy.cfg.j2
+++ b/src/pybind/mgr/cephadm/templates/services/ingress/haproxy.cfg.j2
@@ -74,7 +74,7 @@ backend backend
     balance static-rr
     option httpchk HEAD / HTTP/1.0
     {% for server in servers %}
-    server {{ server.name }} {{ server.ip }}:{{ server.port }} check weight 100
+    server {{ server.name }} {{ server.ip }}:{{ server.port }} check weight 100 inter {{ health_check_interval }}
     {% endfor %}
 {% endif %}
 {% if mode == 'tcp' %}

diff --git a/src/pybind/mgr/cephadm/tests/test_services.py b/src/pybind/mgr/cephadm/tests/test_services.py
index 30b121cc37268b342bd57a8185f4119d65c6b5fa..8998d6f238e2a7fb04eca3cc527e5b5af5b4f0b0 100644 (file)
--- a/src/pybind/mgr/cephadm/tests/test_services.py
+++ b/src/pybind/mgr/cephadm/tests/test_services.py
@@ -1854,7 +1854,7 @@ class TestIngressService:
                                 'balance static-rr\n    '
                                 'option httpchk HEAD / HTTP/1.0\n    '
                                 'server '
-                                + haproxy_generated_conf[1][0] + ' 1.2.3.7:80 check weight 100\n'
+                                + haproxy_generated_conf[1][0] + ' 1.2.3.7:80 check weight 100 inter 2s\n'
                         }
                 }
 
@@ -1979,7 +1979,7 @@ class TestIngressService:
                                 'balance static-rr\n    '
                                 'option httpchk HEAD / HTTP/1.0\n    '
                                 'server '
-                                + haproxy_generated_conf[1][0] + ' 1::4:443 check weight 100\n'
+                                + haproxy_generated_conf[1][0] + ' 1::4:443 check weight 100 inter 2s\n'
                         }
                 }
 
@@ -2103,7 +2103,7 @@ class TestIngressService:
                                 'balance static-rr\n    '
                                 'option httpchk HEAD / HTTP/1.0\n    '
                                 'server '
-                                + haproxy_generated_conf[1][0] + ' 1.2.3.7:80 check weight 100\n'
+                                + haproxy_generated_conf[1][0] + ' 1.2.3.7:80 check weight 100 inter 2s\n'
                         }
                 }
 

diff --git a/src/python-common/ceph/deployment/service_spec.py b/src/python-common/ceph/deployment/service_spec.py
index db7beac4dc22e3ab093ac5ad30bffdf5a8c91fa3..7edd14ce28e5ce56ce0e0fdad68d24da58a5a149 100644 (file)
--- a/src/python-common/ceph/deployment/service_spec.py
+++ b/src/python-common/ceph/deployment/service_spec.py
@@ -1463,6 +1463,7 @@ class IngressSpec(ServiceSpec):
                  extra_container_args: Optional[GeneralArgList] = None,
                  extra_entrypoint_args: Optional[GeneralArgList] = None,
                  custom_configs: Optional[List[CustomConfig]] = None,
+                 health_check_interval: Optional[str] = None,
                  ):
         assert service_type == 'ingress'
 
@@ -1495,6 +1496,8 @@ class IngressSpec(ServiceSpec):
         self.ssl = ssl
         self.keepalive_only = keepalive_only
         self.enable_haproxy_protocol = enable_haproxy_protocol
+        self.health_check_interval = health_check_interval.strip(
+        ) if health_check_interval else None
 
     def get_port_start(self) -> List[int]:
         ports = []
@@ -1525,6 +1528,13 @@ class IngressSpec(ServiceSpec):
         if self.virtual_ip is not None and self.virtual_ips_list is not None:
             raise SpecValidationError(
                 'Cannot add ingress: Single and multiple virtual IPs specified')
+        if self.health_check_interval:
+            valid_units = ['s', 'm', 'h']
+            m = re.search(rf"^(\d+)({'|'.join(valid_units)})$", self.health_check_interval)
+            if not m:
+                raise SpecValidationError(
+                    f'Cannot add ingress: Invalid health_check_interval specified. '
+                    f'Valid units are: {valid_units}')
 
 
 yaml.add_representer(IngressSpec, ServiceSpec.yaml_representer)