Investigate why `reconfigure` and `hup` cause downtime on multi-node Geo deployments
In #225684 (closed), we discovered that in the course of a geo HA upgrade, the reconfigure
and hup puma
steps are causing downtime. As far as I know this issue has not been reported in non-geo setups.
Definition of Done
-
Determine why reconfigure
causes an app restart in geo -
Determine why hup puma
causes an app restart in geo -
Provide steps to mitigate or fix.
Background
This behavior was observed in upgrading %12.10, however the code that causes the reboot seems to also exist in %13.0 (since reconfigure and hup were run on an upgraded node.
See #225684 (closed) for more details.
CC: @jennielouie
Edited by Jennifer Louie