[ovs-dev] [PATCH] ovs-vsctl: Try connecting only once for active connections by default.

Ansis Atteka aatteka at nicira.com
Mon Mar 18 17:32:21 UTC 2013


On Mon, Mar 18, 2013 at 10:26 AM, Ben Pfaff <blp at nicira.com> wrote:
> On Fri, Mar 15, 2013 at 04:13:37PM -0700, Ansis Atteka wrote:
>> I saw few other occurrences where ovs-vsctl is being called with
>> --timeout, Though, I am not sure how safe or necessary it would be to
>> remove --timeout everywhere at this time.
>
> It seems entirely safe to me when --no-wait is in use or the command is
> just a query (that is, it does not modify the database).  Without
> --no-wait, a command that modifies the database can still hang
> "forever" if ovs-vswitchd is not running.  I tried to be selective and
> remove the timeouts only in the former case.
>
> We can probably write another patch to make ovs-vsctl detect if
> ovs-vswitchd is not running and not wait for it to reconfigure in that
> case even without --no-wait.
>
>> Also I am wondering, if someone depended on default behaviour
>> (without --retry flag) to be "retry forever". We will see this
>> soon. Probably libvirt will be able to make use of this "retry once"
>> approach.
>
> It is definitely a change in behavior.  I hope that it does not break
> anything, but it is possible.  If it causes serious problems, we can
> revert it.  Still, I think that it is worth the risk because the new
> behavior is IMO less surprising than the old behavior.
>
>> Besides those 3 small things I already commented this patch looks good to me!
>
> I guess there aren't any followup patches needed after all.  Please
> let me know if you spot anything wrong.

I don't think that any additional patches are required here. Assuming
that you fixed those 3 small things.



More information about the dev mailing list