Many to one where the one can be one of two different fields - postgresql

Question

I've had a look at similar posts on one to many and many to one relationships but I just can't find what I'm looking for. I have 2 tables. The first (vm) contains details of virtual machines. The vms are clustered in pairs and the second table (cluster) has details of the cluster. The set up is as follows:
vm table cluster table
vmId clusterId
ipaddress primaryId
macaddress secondaryId
The relationship is that the vm.vmId can be a match to either cluster.primaryId or cluster.secondaryId. With only an ipaddress I want to be able to get the full cluster details as follows:
clusterid vmId ipaddress macaddress
c1 v1 10.0.10.10 AA:BB:CC:DD:10:11
C1 v2 10.0.10.11 AA:BB:CC:DD:10:12
and that's my problem. With just the ipaddress I can find the vmId and with that I can get the cluster, but I can't then work out how to use the non matching id in the cluster table to get the other vm details. I've tried variables and unions etc., but I must be doing something wrong because I keep coming up blank. I'm looking for a single select statement that will provide the required result. Can anyone suggest a solution please?

Answer 1

First find the cluster(s) with given IP:
select *
from cluster
where exists (
select 1
from vm
where
(vm.vmId = cluster.primaryId or vm.vmId = cluster.secondaryId) and
wm.ipaddress = '127.0.0.1');
Then join this with vm table in the same way:
with clusters as (
select *
from cluster
where exists (
select 1
from vm
where
(vm.vmId = cluster.primaryId or vm.vmId = cluster.secondaryId) and
wm.ipaddress = '127.0.0.1'))
select *
from clusters join vm on (vm.vmId = clusters.primaryId or vm.vmId = clusters.secondaryId);

Answer 2

This is probably what you are looking for, but realize that the vmIds will appear twice, once under the primary cluster, and another under the secondary cluster.
SELECT clusterid, vmId, ipaddress, macaddress
FROM vm
JOIN cluster
ON vm.vmId = cluster.primaryId
UNION ALL
SELECT clusterid, vmId, ipaddress, macaddress
FROM vm
JOIN cluster
ON vm.vmId = cluster.secondaryId
The above can be reduced to:
SELECT clusterid, vmId, ipaddress, macaddress
FROM vm
JOIN cluster
ON (vm.vmId = cluster.primaryId OR vm.vmId = cluster.secondaryId)
If you want to filter the result:
SELECT clusterid, vmId, ipaddress, macaddress
FROM vm
JOIN cluster
ON (vm.vmId = cluster.primaryId OR vm.vmId = cluster.secondaryId)
WHERE ipaddress = '10.0.10.10'

Answer 3

With your help I did finally manage to get the answer I was looking for. With a little modification of your suggestions I finally came up with this.
with clusters as (
select clusterId
from cluster
where exists (
select 1
from vm
where
(vm.vmId = cluster.primaryId or vm.vmId = cluster.secondaryId) and
wm.ipaddress = '10.0.10.10'))
select clusterId, vmId, ipaddress, macaddress
from vm
join cluster on (vm.vmId = cluster.primaryId or vm.vmId = cluster.secondaryId)
where cluster.clusterId = (select * from clusters)
Produced exactly the result I was looking for. Thanks everyone for your assistence.