On Sun, May 27, 2012 at 6:03 AM, Timothy Murphy <gayleard at eircom.net> wrote: > >>> > Wait and schedule a downtime window for it. >>> >>> I don't know what a "downtime window" is in this context. >>> I'm either in the same place as the server, or I am not. >> >> Downtime Window: It's when you schedule a specific time to update the >> machine or make repairs to that it needs. > > As I said, that doesn't make sense in my context. The better equivalent is to have the service running on multiple machines in the first place so no one will notice if one breaks for any reason. > Actually, my question was: what is the probability of failure? > If someone tells me a horse has a 90% chance of winning, > and I back it and it falls down, I don't blame my tipster. I've had yum sessions fail (probably mostly from starting them in a freenx session...) but never to the point where yum-complete-transaction did not fix it, so I don't have a good feeling for what is actually happening. However, I wouldn't expect a forced install of any reasonably close glibc version to break anything. > That's not quite fair. > You also said that if glibc "breaks" PAM will fail and so ssh will fail. > That is certainly a possibility I had not considered. I'd say those are actually more likely than breaking the kernel, but what is your plan if the drives or motherboard fails? Worst case software wise won't be any worse than that. -- Les Mikesell lesmikesell at gmail.com