92 lines
4.0 KiB
Plaintext
92 lines
4.0 KiB
Plaintext
{
|
|
"comments": [
|
|
{
|
|
"unresolved": false,
|
|
"key": {
|
|
"uuid": "ccf9b5b3_3b42a94a",
|
|
"filename": "/PATCHSET_LEVEL",
|
|
"patchSetId": 9
|
|
},
|
|
"lineNbr": 0,
|
|
"author": {
|
|
"id": 28223
|
|
},
|
|
"writtenOn": "2022-02-10T15:37:08Z",
|
|
"side": 1,
|
|
"message": "Sorry - complete noob here on that topic, I\u0027ll get it eventually, but you have to explain it slowly 😊",
|
|
"revId": "8183e63751066ad30320d45668c3cf8a0c706007",
|
|
"serverId": "4a232e18-c5a9-48ee-94c0-e04e7cca6543"
|
|
},
|
|
{
|
|
"unresolved": true,
|
|
"key": {
|
|
"uuid": "b27232c2_f879a07c",
|
|
"filename": "specs/yoga/tripleo_ceph_ingress.rst",
|
|
"patchSetId": 9
|
|
},
|
|
"lineNbr": 114,
|
|
"author": {
|
|
"id": 28223
|
|
},
|
|
"writtenOn": "2022-02-10T15:37:08Z",
|
|
"side": 1,
|
|
"message": "nit: \"is\" ?\nSo, if I get it right:\nthe ingress daemon comes around a ceph-nfs instance managed by cephadm; it opens 2 new ports.\nWhere are those ports exposed? if I get it right, they are only on the internal_api - correct? And HAProxy does the link using tcp connection, and is the actual endpoint services will hit.. ?",
|
|
"range": {
|
|
"startLine": 114,
|
|
"startChar": 60,
|
|
"endLine": 114,
|
|
"endChar": 64
|
|
},
|
|
"revId": "8183e63751066ad30320d45668c3cf8a0c706007",
|
|
"serverId": "4a232e18-c5a9-48ee-94c0-e04e7cca6543"
|
|
},
|
|
{
|
|
"unresolved": true,
|
|
"key": {
|
|
"uuid": "57b7c412_8aa452d8",
|
|
"filename": "specs/yoga/tripleo_ceph_ingress.rst",
|
|
"patchSetId": 9
|
|
},
|
|
"lineNbr": 114,
|
|
"author": {
|
|
"id": 25402
|
|
},
|
|
"writtenOn": "2022-02-10T16:43:49Z",
|
|
"side": 1,
|
|
"message": "Thanks, fixing the nit in the next PS.\nThe ingress daemon is a component made by 2 containers provided by Ceph: haproxy + keepalived.\nHaproxy requires the ports I mentioned in this PS, and they are exposed on the network where the frontend vip is reserved.\n\nIn particular, if you have something like the following:\n\n- name: Create the ingress Daemon spec definition for nfs\n become: true\n ceph_mkspec:\n service_type: ingress\n cluster: \"{{ tripleo_cephadm_cluster }}\"\n apply: true\n hosts: \"{{ _hosts | unique }}\"\n render_path: \"{{ tripleo_cephadm_spec_home }}\"\n spec:\n backend_service: nfs.nfs\n frontend_port: \"{{ tripleo_cephadm_nfs_ingress_frontend_port }}\"\n monitor_port: \"{{ tripleo_cephadm_nfs_ingress_monitor_port }}\"\n virtual_ip: \"{{ tripleo_cephadm_nfs_ingress_vip | default(\u00270.0.0.0\u0027) }}\"\n virtual_interface_networks: \"{{ tripleo_cephadm_ceph_nfs_address_block }}\"\n\nyou can see in the spec dict we have both the VIP and the associated network (it can be a composable network, like StorageNFS).\nNetworks and vips are already provisioned in this process after metal (networkv2 takes care about the networking definition/creation).\nAt this stage we don\u0027t have any firewall rule applied, as it will be done during the overcloud deployment (like any other Ceph component).",
|
|
"parentUuid": "b27232c2_f879a07c",
|
|
"range": {
|
|
"startLine": 114,
|
|
"startChar": 60,
|
|
"endLine": 114,
|
|
"endChar": 64
|
|
},
|
|
"revId": "8183e63751066ad30320d45668c3cf8a0c706007",
|
|
"serverId": "4a232e18-c5a9-48ee-94c0-e04e7cca6543"
|
|
},
|
|
{
|
|
"unresolved": false,
|
|
"key": {
|
|
"uuid": "77cb80f5_ca970f61",
|
|
"filename": "specs/yoga/tripleo_ceph_ingress.rst",
|
|
"patchSetId": 9
|
|
},
|
|
"lineNbr": 114,
|
|
"author": {
|
|
"id": 28223
|
|
},
|
|
"writtenOn": "2022-02-11T06:39:26Z",
|
|
"side": 1,
|
|
"message": "Ack",
|
|
"parentUuid": "57b7c412_8aa452d8",
|
|
"range": {
|
|
"startLine": 114,
|
|
"startChar": 60,
|
|
"endLine": 114,
|
|
"endChar": 64
|
|
},
|
|
"revId": "8183e63751066ad30320d45668c3cf8a0c706007",
|
|
"serverId": "4a232e18-c5a9-48ee-94c0-e04e7cca6543"
|
|
}
|
|
]
|
|
} |