More resilient startup
Currently when promscale cannot connect to database it almost immediately crashes. In environments where database is started at the same time as promscale (like tobs) this can lead to multiple Crash Loops. Since kubernetes does an exponential crashloop backoff, it leads to long delay in startup.
My proposal is to add a timeout at the startup to check for connection for about 120s (ideally this should be customizable via cli flag). This way long database startup won't affect the whole stack startup that much.
pinging @VineethReddy02 @cevian for task prioritization after release.
duplicate to #592
closing this issue, as the same issue is created by the user in #592