On 08/28/2015 10:13 AM, Johnny Hughes wrote: > On 08/28/2015 08:43 AM, Robert Moskowitz wrote: >> >> On 08/28/2015 09:22 AM, Johnny Hughes wrote: >> >>> But we have to stabilize what we want in os/ (ie, decide on a full >>> armv7hl package set and get it built in the 7.1 release tree) .. THEN do >>> updates/ ... THEN add extra packages like epel or sernet rpms or >>> whatever. Otherwise, we are never going to be able to release this >>> stuff. >> Johnny, >> >> I am all for staying off the Internet and not doing extras. >> >> My box is on a lab subnet behind 2 levels of firewalls and NATs. SHOULD >> be safe; except from myself! >> >> Not much I know how to test. I CAN build some mariadb stuff and try >> that out, but then I need some epel tools. Well maybe not, if I copy >> the production db over. >> >> BIND is 'easy', provided I have a Cubieboard2 image. Nothing there but >> bind and httpd. >> >> sernet gets me testing a number of different things, and it DOES seem >> that really all I need extra is in those sernet rpms that Nicolas was so >> kind to build. I can stay at what I have for some time. >> >> I did back off on epel once i realized I was barking up the wrong tree; >> and you are right that we (ie the developers) need to get more of the >> basics available. I am booting with the original kernel, as the kernel >> update did not work, and I am not going to spend time learning how to >> fix this. >> > Right .. we also need to know the steps involved in fixing the kernel > updates, so we can then put those in a post and postun step in the kenel > SRPM. If we can document every step to make the updates happen via the > command line then I should be able to take the commands and come up with > a script. This would be a huge help as well. I am no help. For Fedora-arm, I just use their installer. > I know rpi2 uses a proprietary kernel. We would need a way to test for > that and NOT run the script to make those machines get an updated kernel > in the /boot/ partition, etc. One of the reasons I will not go with rpi. > WRT new images, one can rebuild an image from rbf and use the new kernel > for cubietruck and the other boards that actually use the CentOS kernel > and not their own kernel. Obviously if you use the entire new image, you > loose your current setup. Another option would be to sync the /boot/ > partition from the new kernel over the /boot/ partition on the machine > after installing the rpm. (I have not tested that yet, but I think it > would work for things that use the CentOS kernel) For now, I have no problems with trashing what I have and starting with a new mSD card layout. I have documented what I have done and can redo it without the missteps. I will not be doing any rbf building. > >> You get me images, I will do my best to test. > BTW, I didn't want to sound harsh before and I do want to get this done. > But I don't want anyone getting hacked because we are currently doing > updates, etc. So wanted the warning to go out, etc. :) Hey, I have been reading your posts on the Centos list for how long have you been working on it? :)