Hmm.. there are a *lot* of run_program() calls in sysinst which do not check the return value.
Would be a very useful task for someone to run through them all and add checks Anyone interested? :) On 16 November 2014 12:49, David Brownlee <[email protected]> wrote: > It was using more than 640M of VM? > > While I agree not having a failure reported as success would be the > top priority, the thought that it was using that much VM during > install is... scary. > > What arch is this on? i386 has a set_swap() in sysinst to use > swapspace if less than 32M of ram, but amd64 does not... I suspect > that check should be made MI... > > > On 15 November 2014 16:15, Andy Ruhl <[email protected]> wrote: >> I've attached 3 screenshots. During each attempt, I was installing in a >> virtual machine. The first (that I didn't document) was a Linux KVM. This >> time was on VMWare Fusion (with screenshots). Each time I set it up with 128 >> megs of memory, 512 megs of swap, and a 15 gig disk. >> >> The concern is not necessarily that the pkgin config process failed, it's >> that it was marked as completed afterward. >> >> Andy >> >> On Fri, Nov 14, 2014 at 3:37 PM, Andy Ruhl <[email protected]> wrote: >>> >>> Hello all. >>> >>> I just had something happen that I didn't capture but I will try to at >>> some point. >>> >>> This is when trying to install pkgin using the install kernel just after >>> the full install of sets is finished. >>> >>> During the "building database" phase, it issued a message something like >>> "killed, ran out of swap". >>> >>> But the process ended and returned to the menu. The process was marked as >>> "DONE". >>> >>> Seems like it should be marked failed if the process fails? >>> >>> I searched a bit and I see a few PRs but nothing specific to pkgin failing >>> and then getting marked "DONE". Is this a bug? >>> >>> Andy >> >>
