If you use fleet, you can leverage systemd's own requirements semantics. If you use kubernetes, you can buffer delays for each component container or rely on the fact that kubernetes will restart failed containers. Just make sure your containers exit on failure.
On Thu, Jul 14, 2016 at 8:15 PM Derek Mahar <derek.ma...@gmail.com> wrote: > What is the recommended approach to synchronizing containers in CoreOS? > For example, how might CoreOS ensure the following container execution > order at startup? > > 1. Start PostgreSQL Server > 2. Create schema (using separate one-shot Liquibase container) > 3. Launch application > > The application cannot run without its database schema and Liquibase > cannot create the database schema without PostgreSQL Server being ready to > receive SQL statements. Which mechanisms in CoreOS can we use to guarantee > this ordering? > > Thank you, > > Derek > -- Seán C McCord CyCore Systems, Inc +1 888 240 0308 PGP/GPG: http://cycoresys.com/scm.asc
