> On 20 Oct 2017, at 14:06, Jason Lixfeld <[email protected]> wrote: > > Hello! > > For those running (a lot of?) ME3600s (24FS) in your network - I’m probing to > see what your views are on the current state of software stability on those > boxes. > > I’ve got a couple hundred of these in the field, and it took almost 4 years > before the code was stable, so when we found that stable version, we latched > onto it out of sheer terror of going through another few years of hell, and > as such, we’ve been on that code for almost the last 4 years. > > I need to be on at least 15.4(3)S1 to resolve all the bugs on my current > list, and while reviewing the ME3600 specific release notes for 15.4S and > 15.5S (http://tinyurl.com/yc9uudvz), it seems to me that generally speaking, > the 15.5S train as a whole has far, far, FAR fewer feature additions than the > 15.4S train. That seems to suggest to me that 15.5S is focusing heavily on > bug fixes than features, so 15.5S would likely be the obvious choice. > > What’s the collective's experience around this? > > Thanks in advance!
Can’t say things seem any more stable with the newer releases as we just tried to start a program to upgrade to the recommended 15.5S and immediately hit an issue with npm_main process pegging the CPU cycles if you have copper SFP’s installed in any state other than UP. This issue is also present in the latest 15.6S. This occurs with both Cisco OEM and Flexoptics SFP’s across various types (GLC-T and GLC-TE). TAC case open and escalated to BU, but this has put all thoughts on upgrades on hold for us until this is resolved. _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
