(re-adding list as I accidentally cut it off previously) Thanks Baptiste, that's perfectly clear.
Is 'change the config and reload' the best way to do manual failover? Or is there some kind of signal I can send haproxy to make it 'flip' between backends? On Fri, Jan 8, 2016 at 7:20 AM, Baptiste <[email protected]> wrote: > Hi Mike, > > > > Regarding my setup, I'd like to have a blue/green deployment type setup - > > eg, only one environment active, with manual switching between them. > > > > - If A is active, and A fails, do not do anything, even if B is up. > > well, this is not possible with the stick-table example. > > > - If A is active, and B is made active, traffic keeps going to A. > > > The setup in the article seems to do what I want, but I understand > > 'stickiness' is only for existing clients. Is there a way for *new* > clients > > to also only go to one particular server at a time? > > For both statements above, you simply need a default_backend option > and change the value pointed by it in your frontend then reload > HAProxy. > > Baptiste >
