Could folks cc the bugs to the list so we can track. Ideally we would like to leverage our images in upstream k8s e2e tests and examples.
Cheers, Tim ----- Original Message ----- > From: "Stephen C. Tweedie" <[email protected]> > To: "Colin Walters" <[email protected]> > Cc: [email protected] > Sent: Monday, May 18, 2015 9:39:50 AM > Subject: Re: [atomic-devel] Base image sizes > > Hi, > > On Fri, 2015-05-15 at 16:41 -0400, Colin Walters wrote: > > On Fri, May 15, 2015, at 01:11 PM, Matthew Miller wrote: > > > > > In seriousness, this has crept up quite a lot since the initial > > > versions we'd worked on a while ago, and has some huge room for > > > improvement. I see it includes dracut and grub2, > > > > I think this an Anaconda regression. Using `bootloader --location=none` > > and `%packages --nocore` was supposed to avoid installing a bootloader, > > but we're ending up with one again. > > > > If you look at /var/log/anaconda/anaconda.log in the image (yes, > > log files are in the image by default), you'll see: > > > > 18:28:07,936 INFO anaconda: bootloader GRUB2 on X86 platform > > Right, I noticed grub2 in the images this morning. Removing that (grub2 > plus grub2-tools) frees up over 50MB; over 70MB once you remove deps > that only grub was using. > > > I'll file a bug with Anaconda. > > Thanks, sounds like this is a big win. > > --Stephen > > > -- Cheers, Timothy St. Clair Red Hat Inc.
