Hi there, I recently deployed a load balancer on our openstack private cloud. I used this manual - https://docs.openstack.org/octavia/latest/user/guides/basic-cookbook.html to create the load balancer. However, after creating and trying to access it, it returns an error message saying "No server is available to handle this request". Also on the dashboard, "Operating status" shows offline but "provisioning status" shows active. I have two web applications as members of the load balancer and I can individually access those web applications. Could someone please point me in the right direction? Thanks.
On Mon, Apr 5, 2021 at 4:14 PM Adekunbi Adewojo <aadewojo@gmail.com> wrote:
Hi there,
I recently deployed a load balancer on our openstack private cloud. I used this manual - https://docs.openstack.org/octavia/latest/user/guides/basic-cookbook.html to create the load balancer. However, after creating and trying to access it, it returns an error message saying "No server is available to handle this request". Also on the dashboard, "Operating status" shows offline but "provisioning status" shows active. I have two web applications as members of the load balancer and I can individually access those web applications.
Hi, provisioning status ACTIVE shows the load balancer was successfully created but operating status OFFLINE indicates that the amphora is unable to communicate with the Octavia health-manager service. Basically, the amphora should report its status to the hm service using UDP messages (on a controller, you can dump the UDP packets on the o-hm0 interface), do you see any errors in the hm logs? I would recommend enabling the debug messages in the health-manager, and to check the logs, you should see messages about incoming packets: Apr 06 10:02:20 devstack2 octavia-health-manager[1747820]: DEBUG octavia.amphorae.drivers.health.heartbeat_udp [-] Received packet from ('192.168.0.73', 11273) {{(pid=1747857) dorecv /opt/stack/octavia/octavia/amphorae/drivers/health/heartbeat_udp.py:95}} The "No server is available to handle this request" response from the amphora indicates that haproxy is correctly running on the VIP interface but it doesn't have any member servers. Perhaps fixing the health-manager issue will help to understand why the traffic is not dispatched to the members.
Could someone please point me in the right direction?
Thanks.
Thank you for your response. However, I do not know how to enable debug messages for the health monitor. I do not even know how to access the health monitor log because I can't ssh into the load balancer. On Tue, Apr 6, 2021 at 11:09 AM Gregory Thiemonge <gthiemonge@redhat.com> wrote:
On Mon, Apr 5, 2021 at 4:14 PM Adekunbi Adewojo <aadewojo@gmail.com> wrote:
Hi there,
I recently deployed a load balancer on our openstack private cloud. I used this manual - https://docs.openstack.org/octavia/latest/user/guides/basic-cookbook.html to create the load balancer. However, after creating and trying to access it, it returns an error message saying "No server is available to handle this request". Also on the dashboard, "Operating status" shows offline but "provisioning status" shows active. I have two web applications as members of the load balancer and I can individually access those web applications.
Hi,
provisioning status ACTIVE shows the load balancer was successfully created but operating status OFFLINE indicates that the amphora is unable to communicate with the Octavia health-manager service. Basically, the amphora should report its status to the hm service using UDP messages (on a controller, you can dump the UDP packets on the o-hm0 interface), do you see any errors in the hm logs? I would recommend enabling the debug messages in the health-manager, and to check the logs, you should see messages about incoming packets:
Apr 06 10:02:20 devstack2 octavia-health-manager[1747820]: DEBUG octavia.amphorae.drivers.health.heartbeat_udp [-] Received packet from ('192.168.0.73', 11273) {{(pid=1747857) dorecv /opt/stack/octavia/octavia/amphorae/drivers/health/heartbeat_udp.py:95}}
The "No server is available to handle this request" response from the amphora indicates that haproxy is correctly running on the VIP interface but it doesn't have any member servers. Perhaps fixing the health-manager issue will help to understand why the traffic is not dispatched to the members.
Could someone please point me in the right direction?
Thanks.
I also suggest to verify if the load balancer network used by amphora ave access to controllers port udp 5555. If you want to access the load balancers instances, you can specify in the configuration : amp_ssh_key_name = your ssh key So you can logon on loadbalancer instances via ssh Il giorno mar 6 apr 2021 alle ore 17:40 Adekunbi Adewojo <aadewojo@gmail.com> ha scritto:
Thank you for your response. However, I do not know how to enable debug messages for the health monitor. I do not even know how to access the health monitor log because I can't ssh into the load balancer.
On Tue, Apr 6, 2021 at 11:09 AM Gregory Thiemonge <gthiemonge@redhat.com> wrote:
On Mon, Apr 5, 2021 at 4:14 PM Adekunbi Adewojo <aadewojo@gmail.com> wrote:
Hi there,
I recently deployed a load balancer on our openstack private cloud. I used this manual - https://docs.openstack.org/octavia/latest/user/guides/basic-cookbook.html to create the load balancer. However, after creating and trying to access it, it returns an error message saying "No server is available to handle this request". Also on the dashboard, "Operating status" shows offline but "provisioning status" shows active. I have two web applications as members of the load balancer and I can individually access those web applications.
Hi,
provisioning status ACTIVE shows the load balancer was successfully created but operating status OFFLINE indicates that the amphora is unable to communicate with the Octavia health-manager service. Basically, the amphora should report its status to the hm service using UDP messages (on a controller, you can dump the UDP packets on the o-hm0 interface), do you see any errors in the hm logs? I would recommend enabling the debug messages in the health-manager, and to check the logs, you should see messages about incoming packets:
Apr 06 10:02:20 devstack2 octavia-health-manager[1747820]: DEBUG octavia.amphorae.drivers.health.heartbeat_udp [-] Received packet from ('192.168.0.73', 11273) {{(pid=1747857) dorecv /opt/stack/octavia/octavia/amphorae/drivers/health/heartbeat_udp.py:95}}
The "No server is available to handle this request" response from the amphora indicates that haproxy is correctly running on the VIP interface but it doesn't have any member servers. Perhaps fixing the health-manager issue will help to understand why the traffic is not dispatched to the members.
Could someone please point me in the right direction?
Thanks.
participants (3)
-
Adekunbi Adewojo
-
Gregory Thiemonge
-
Ignazio Cassano