Nuxeo Server won't start

I'm revising this ticket because 1) I have seen the problem now on localhost, AWS, and GCP, so it's not GCP-specific 2) I have seen the problem with many different plug-ins so I no longer think it's nuxeo-retention-specific.

The generic problem is that Nuxeo Server fails to start with "Starting process is taking too long - giving up."

It usually happens when a stack has "lots" of plug-ins but I have not been able to isolate it to a certain set of plug-ins.

In the original ticket it happened when adding nuxeo-retention. In the latest iteration it happens when adding pii-redaction (it's a multi-layer Studio project).

In most cases if I remove the most recently added plug-in, it works ok. But the strange thing is sometimes after iterating over troubleshooting, adding and removing plug-ins, it will sometimes work fine after a while with all plug-ins installed. I.e. it seems to be related to 1) the quantity of plug-ins 2) the order in which they are deployed?

This time I have some thread dumps from a failing server so I will attach those.