[ec2-beta] init level 2-4 + network access

[Darren Govoni]

Yeah, good point. After looking at your perl snippet I understand what
its doing now. Thanks again.

On Mon, 2009-03-30 at 18:17 -0700, Eric Hammond wrote:
> Darren:
> 
> A sleep simply reduces the risk of failure but does not eliminate it.
> Plus, you may be waiting longer than necessary in many cases.
> 
> Why not explicitly wait for the network to come up so that you both
> eliminate the risk and speed up the process?
> 
> --
> Eric Hammond
> ehammond at thinksome.com
> 
> 
> 
> Darren Govoni wrote:
> > Eric,
> >    That's an interesting idea about the boot speed (which is lightening
> > fast). Let me see about that, maybe putting a sleep in my script above
> > the svn call. I went back to beta1 since, but really need to keep up
> > with the program.
> > 
> > I'll report back and thank you again for the tip.
> > 
> > Darren
> > 
> > On Sun, 2009-03-29 at 03:25 -0700, Eric Hammond wrote:
> >> Darren:
> >>
> >> Did you ever solve this issue?
> >>
> >> If your scripts are being run from /etc/rc*.d/ (or symlinks there) then
> >> they are run as root.  The order in which they are run relative to
> >> network setup depends partly on the names of the files or symlinks in
> >> that directory and partly on how fast DHCP resolves.
> >>
> >> In the Ubuntu images I publish on http://alestic.com I found they booted
> >> so fast that I needed to put in some checks to wait for the network to
> >> come up (DHCP).  I did this using a command like the following (only
> >> works on EC2):
> >>
> >>   perl -MIO::Socket::INET -e '
> >>    until(new IO::Socket::INET("169.254.169.254:80")){sleep 1}'
> >>
> >> I have a suggested ticket open in launchpad for the official images:
> >>
> >>   https://bugs.launchpad.net/ubuntu-on-ec2/+bug/308530
> >>
> >> Each process requiring network will need to implement its own check.
> >>
> >> --
> >> Eric Hammond
> >> ehammond at thinksome.com
> >>
> >>
> >>
> >> Darren Govoni wrote:
> >>> Hi,
> >>>   In beta 1 I had some init scripts I wrote that pulled updates from a
> >>> subversion server outside of EC2. In beta2 those scripts don't quite
> >>> work entirely. After adding some logging, they seem to get hung or bail
> >>> out on svn calls that attempt network access. That is, they run up to
> >>> that point.
> >>>
> >>> So my question is, is there any network constraints at init 2 that would
> >>> prevent my script from working?
> >>>
> >>> I will dare to also ask if they run as root like before, which I think
> >>> my scripts require at boot time.
> >>>
> >>> Warm regards,
> >>> Darren
> >>>
> >>>
> >