Originally posted by: alpharob1
I've had a number of issues with Entitlement downloads and/or general installation problems. Problems that should not be. I have several examples and recent problems have me even more concerned. I don't want to dig through emails to try and detail every one of them. My point it trotting this out here is two-fold. First, to point out that simple testing is not being performed prior to sticking ISOs out there. Secondly, to try to encourage a raising of the game. I want to speak to three specific examples. First, in December 2012 (yes, I'm going back for this one, but this one is so brain dead as to be boggling) I downloaded AIX 6.1 TL8 SP0 ISOs from Entitlement. Used it to perform an update. bos.aso was missing from that distribution and left me with a bunch of broken filesets as bos.aso was a pre-req for quite a few. I figured my way out ot that. A few first thoughts were along the lines of "how did this even go out the door?"
Recently, I've had a lot of not-so-fun with VIO 2.2.3.0 distro. As we've seen a recent flash with FC 5899 (4 port 1 gig card) , you can't create Etherchannel aggregate nor NIB (AIX 6.1 TL9 and 2.2.3.0 , by the way). I then applied 2.2.3.1, same thing. At the time, google lead me here: http://www-01.ibm.com/support/docview.wss?uid=isg1IV52695 , which in turn pointed here: http://www-01.ibm.com/support/docview.wss?uid=isg1fixinfo145795 that link talked about a fix , but wasn't pointing to anything. A co-worker told me when he ran into this , a call to IBM and was sent "IV52695s1a.131216.epkg.Z" which fixed it. I had no time to chase and chase so a way out for me, was to drop back to 2.2.2.2 and go to 2.2.2.4, SCA-REW 2.2.3.0. Thoughts with this are how does a long stable driver get broken (FC 5899 has been out quite a while)? Why wouldn't a common operation (LACP and NIB) not be tested prior to stamping a distro? Why links that point no where?
That same 2.2.3.0 distro caused me additional pain and delays. While installing that same DVD on a p460 a week later (10 gbit infra, maybe 10 gbit isn't broken, client request of 2.2.3.x.. nervousness washes over me) , I was getting a 0518 on boot and a hang. 0518 in the googleples shows "mount /usr failed" - this is the same DVD that a week earlier booted (but I couldn't create a LACP group, nor NIB). So thinking maybe I had some kind of corruption, downloaded 2.2.3.0 again, burned new DVDs, booted ... wait for it ... same error 0518 and a hang on the p460. Way out of the forest here was to use the trusty 2.2.2.2 and work my way to 2.2.3.x. First thought here is: "This freakin' 2.2.3.0 build was never booted on a p460"
Questions:
- What kind of regression testing is being performed prior to stamping an ISO? Is that test suite constantly growing? If not, why not?
- Are all current and common configurations re: hardware and typical implementations being tested prior to stamping an ISO? If not, why not?
I find this rather annoying and embarrassing to spend hours in front of clients trying to perform common operations, losing my mind wondering if I am not doing something correctly (age) and then coming to the conclusion "oh, another distro issue... whew, I thought I was losing it!"
Rob