Closed diegoakechi closed 3 years ago
This is apparently a regression of https://github.com/SUSE/ha-sap-terraform-deployments/pull/710
This is apparently a regression of #710
my bad ... I will look into it.
@yeoldegrove No worries ;)
A decision has to be made:
/etc/sap_host_exporter/${SID}_PAS${INSTANCE}.yaml
and this comes directly from the pillarThere might be other solutions that I cannot think of at the moment.
@arbulu89 What is your opinion on this?
HI @yeoldegrove ,
Assuming that the PAS and AAS services won't move from their host machines, I think we could use the option 2, as it looks the easier and quite reasonable.
summed up:
The Netweaver virtual IPs for AS/ERS are reachable via different mechanism for public clouds:
azure -> load-balancer gcp -> route to VM (changed by cluster) aws -> route to VM (changed by cluster) For the Netweaver virtual IPs for PAS/AAS there is currently no mechanism implemented to make them reachable to other nodes.
This PR makes these PAS/AAS virtual IPs reachable via these mechanisms:
azure -> vIP on network interface on VMs load-balancer setup would also be possible in theory but is way more complex (e.g. create rules per node) gcp -> route to VM (static) aws -> route to VM (static)
Used cloud platform Azure
Used SLES4SAP version SLES for SAP 15.2
Used client machine OS openSUSE TW.
Expected behaviour vs observed behaviour
The PAS and AAS sap_host_exporter targets should be available on Prometheus, but currently, they appear configured by Down.
Currently, there is a secondary IP and vHostname
sapha1pas
used by the sap_host_exporter monitoring. Onetc/hosts
we see:But there is no Static IP associated with the NIC used by the machine. Adding it manually solves the problem (IP 10.74.1.36 was added manually):
How to reproduce
The default deployment of HANA + Netweaver + Monitoring using the Develop branch produces this end-result.
Logs This was spot on the Trento Demo deployment. The logs can be collected there.