This is reverted for now. The trick is we need to pass the current Runner to the function in the fork.
I'm going to investigate some options about how to do this though. On Fri, Feb 7, 2014 at 5:42 PM, Michael DeHaan <[email protected]> wrote: > This currently doesn't allocate a new Runner between tasks. > > I am investigating. > > > > > On Fri, Feb 7, 2014 at 5:06 PM, Michael DeHaan <[email protected]>wrote: > >> One of the previous aspects of Ansible was that if you were using a high >> fork count (such as 100-400) there was a performance cost to allocating and >> combining the forks. This cost was not trivial and is actually where >> Ansible could spend most of it's CPU time. >> >> I just committed some code that pre-allocates the series of forks ansible >> uses, so that in a playbook they are not allocated at every task step. >> >> The result of this is playbooks should run blindingly faster now. >> >> I'd appreciate testing and bug reports in case we've broken something >> subtle. >> >> In particular: >> >> * Control-C handling is different, and may not yet be perfect - want to >> make sure we don't leave zombies around and do kill any outstanding ops >> >> * I haven't super-well-tested things with the prompts for host keys on >> new hosts when host key checking is enabled. Should work like before. >> >> This should be included in 1.5 and is currently available on the devel >> branch. >> >> Let me know how it works for you! >> >> Thanks! >> >> --Michael >> > > -- You received this message because you are subscribed to the Google Groups "Ansible Project" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. For more options, visit https://groups.google.com/groups/opt_out.
