Bug #46698
openRadosGW - Swift + loadbalancer
0%
Description
I try to configure a simple rados gateway with three nodes and I add a virtual IP address for the load balancer. I make calls to the gateway using S3 and SWIFT.
In the case of S3 everything works, I can make requests directly on the node where there is a gateway or use the virtual address and it responds well.
In the case of SWIFT, I can query the nodes and it's fine, but when I use the virtual IP address it doesn't respond. The call sequence is the same for the nodes and the virtual IP address.
Version Info
ceph version : 15.2.4 (7447c15c6ff58d7fce91843b705a268a1917325c) octopus (stable)
OS : CentOS 8
Host file (ceph-ansible)
...
[rgws]
node-01
node-02
node-03
[rgwloadbalancers]
node-01
node-02
node-03
...
- GENERAL #
###########
haproxy_frontend_port: 80
haproxy_frontend_ssl_port: 443
haproxy_frontend_ssl_certificate:
haproxy_ssl_dh_param: 4096
haproxy_ssl_ciphers:
- EECDH+AESGCM
- EDH+AESGCM
haproxy_ssl_options:
- no-sslv3
- no-tlsv10
- no-tlsv11
- no-tls-tickets #
virtual_ips:
- XXX.XXX.XXX.189 #
virtual_ip_netmask: 24
virtual_ip_interface: eno5
/etc/ceph/ceph.conf (node1 example)
[client.node1.rgw0]
host = node1
keyring = /var/lib/ceph/radosgw/ceph-rgw. node1.rgw0/keyring
log file = /var/log/ceph/ceph-rgw-node1.rgw0.log
rgw frontends = beast endpoint=xxx.xxx.xxx.190:8080 rgw thread pool size = 512
[client.rgw.node2.rgw0]
host = node2
keyring = /var/lib/ceph/radosgw/ceph-rgw.node2.rgw0/keyring
log file = /var/log/ceph/ceph-rgw- node2.rgw0.log
rgw frontends = beast endpoint=xxx.xxx.xxx.191:8080 rgw thread pool size = 512
[client.rgw.node3.rgw0]
host = node3
keyring = /var/lib/ceph/radosgw/ceph-rgw.node3.rgw0/keyring
log file = /var/log/ceph/ceph-rgw-node3.rgw0.log
rgw frontends = beast endpoint=xxx.xxx.xxx.192:8080 rgw thread pool size = 512