We've just updated VRLI in our lab environment from v4.8 to v8.0 but encountered a couple of issues along the way. Am documenting them here for the benefit of others and any input from VMware.
Failure to migrate network adapter gateway configuration
The gateway for the network adapter wasn't correctly migrated during the conversion to Photon OS. This resulted in the upgrade appearing to hang as the appliance was unable to communicate with any system outside of its IP subnet. The issue is alluded to in the Release Notes but they're somewhat vague and lack instructions on how to check if you're affected and any workaround steps:
The issue is caused by the /etc/sysconfig/network/ifcfg-eth0 file "missing" a GATEWAY='w.x.y.z.' statement. I say "missing" as this doesn't appear to be where SLES typically saves static routes. Those are contained in the /etc/sysconfig/network/routes file, which is what the upgrade process should presumably be looking at. In our case it contained a single line specifying the default route.
As a workaround, you can SSH to the box and add the line the upgrade scripts are expecting before performing the upgrade.
Regression in handling of STARTTLS for SMTP
After upgrading we had to modify the SMTP configuration to not use STARTTLS. This was working fine previously but now sending a test email with STARTTLS enabled fails with the error:
The issue appears to be something in the Java configuration of VRLI. In our configuration we're sending to Office 365 via "Direct Send". Keen to get any input from VMware on this one.
As a workaround you can disable STARTTLS, however, you may wish to do so before starting the upgrade so you don't potentially lose any emails before you can apply the workaround post-upgrade.
Also static routes need te be re-entered again.
For anyone who views this thread there's been some updates since my original post:
Failure to migrate network adapter gateway configuration
While not fixed, there's now a KB article detailing the issue and workaround steps:
https://kb.vmware.com/s/article/76067
Regression in handling of STARTTLS for SMTP
I'm not aware of any workaround or update to this issue.
Hi,
Have you try to open an SR on the issue?
This is a home lab environment with EVALExperience so I have no support entitlement.