Bug#659229: prio_callout is not executed (manpage unclear or outdated?)

[Holger Levsen]

Hi,

On Donnerstag, 9. Februar 2012, Ritesh Raj Sarraf wrote:
> > I'd love to, but I cannot find information how to. Can you please point
> > me in the right direction? Or just give an example for the usecase
> > outlined in this bug report.
> 
> Have a look at the multipath.conf.annotated file in the example/ folder.
> 
> To quote:
> #       # name    : prio
> #       # scope   : multipath & multipathd
> #       # desc    : the default function to call to obtain a path
> #       #           priority value. The ALUA bits in SPC-3 provide an
> #       #           exploitable prio value for example.
> #       # default : const
> #       #
> #       prio    "alua"
> #
> #       #
> #       # name    : prio_args
> #       # scope   : multipath & multipathd
> #       # desc    : The arguments string passed to the prio function
> #       #           Most prio functions do not need arguments. The
> #       #           datacore prioritizer need one.
> #       # default : (null)
> #       #
> #       prio_args       "timeout=1000 preferredsds=foo"
> #
> 
> 
> In case you want to have your own prioritizer, have a look at
> libmultipath/prioritizers/
> You'll have to add your prioritizer and rebuild it. Otherwise, if your
> storage controller is a recent standards compliant array, you can try to
> use ALUA.

that looks way more complicated than the old approach. my users (those 
configuring multipath) tell me they'd prefer to downgrade to 0.4.8 and stick 
with that...

> BTW, what exact use case do you have? I see you've chosen to queue i/o
> when all paths are down. So you'd also want to use queue_if_no_path.


cheers,
	Holger