We have a 10g server being used for development. Recently, as activity has increased, Oracle has begun to refuse connections, and we cannot identify the bottleneck. Some facts about this VM:
- It is assigned 4GB RAM & 2 vCpus (2.0GHz cores)
- This VM does not have any reservations, nor do any VM's on the cluster (it's an ad-hoc Dev environment 50% servers / 50% VDI)
- I fear that setting a reservation will cause the HA admittance algorithm to throw me into a yellow or red state and trigger isolation response for the VM's that are already running
- Its storage is a vmdk file, not an RDM LUN
- It has one virtual network adapter, hosts in cluster have 4-6 nics for virtual machine network
- It is on a DRS-enabled cluster, commonly sharing space with 15+ other virtual machines (Quad core 2GHz / 32 GB RAM PE2950 hosts)
- Protocol analyzer shows a few tcp retransmissions to/from this host
- The VC graphs aren't showing anything alarming, and the guest usually has 1.5GBRAM+ avail.
- No other VM's are having issues
I will be completely honest I do not know Oracle, but if anyone has any suggestions on improving the performance of this VM to the point where I can rule out VMWare as the culprit of the refused connections, that would be great. According to the DBA, Oracle is not throwing any errors and his traces aren't showing anything wrong.
Thanks!