"trade-off between time [developing] and time spent debugging such cases when they do
happen"
Your call. All I know is, it took me over a month to install oVirt, including three weeks
of one-to-one time with Simone Tiraboschi from Red Hat. He sent me eleven emails but
eventually gave up, baffled as me. It shouldn't be this hard. Others who are having
this or similar problems will just abandon oVirt: ("I'm just wondering if I
should cut my losses with oVirt"):
https://lists.ovirt.org/archives/list/users@ovirt.org/thread/PZJYNAKPYNQU...
The biggest challenge is to find the relevant error. What would be useful is a log
aggregator. If oVirt had a journalctl type app running on the host that tails ALL the
logs including the engine logs from hosted-engine (via ssh), everything would be in one
place and easy to spot. Currently, you need fairly detailed knowledge of the architecture
and install process to (i) find the log files (ii) whittle down to the one displaying the
problem. Yes, I know you guys have a log-packaging app that compresses them up, so they
can be sent to Red Hat for inspection (does this even include hosted-engine logs?). But
with a journalling app, users would be able to spot the error themselves and most of the
time (if it's not a new bug), fix it on their own like I did. And yes, I know once
set up, users will have their own log aggregator in the form of Kibana, Splunk, etc but
these don't help during the initial install.