On Thu, 9 Oct 2003, David Boyes wrote: > Date: Thu, 9 Oct 2003 11:24:34 -0400 > From: David Boyes <[EMAIL PROTECTED]> > Reply-To: Linux on 390 Port <[EMAIL PROTECTED]> > To: [EMAIL PROTECTED] > Subject: Re: Updated OCOs for Red Hat 7.2? > > > > Since the network adapter routines are now open-source, why > > not just compile > > > them along with your kernel? > > Supportability. Red Hat has told us that we are unsupported > > if we even so > > much as patch our existing kernel, let alone build a new kernel... > > Interesting. So, they're telling you that if you assemble a new package from > their pieces, including only their modules and no OCO code, that you don't > get support? Even if you supply the source to the necessary modules? > > I would question that response to their management. I can understand their > stance on not supporting kernels including OCO stuff, but in a scenario > where there is clearly no OCO stuff, I think they can do better than that. > If not, then you need a new support vendor. > > -- db >
I've (mostly) moved from RHL mostly because RH has changed the way it deals with customers. I can understand RH not wanting to support software it didn't provide to you _in its standard terms of business_. Refusing to do so on any account seems sheer folly to me. RH provides a heavily-patched kernel, and I quite see that it wouldn't want to waste time soving problems in the standard kernel that are fixed in the kernel RH provides, BUT if the customer, for some reason, wants to use the standard kernel and is prepared to pay an additional fee for its support, why not? If RH gets its costing right, it's dollars in the bank. Also, its non-support of OCO modules is entirely too rigid to my mind. Sure it can't easily support them in the way it supports other software, but IBM can support its OCO stuff, and presumably would be prepared to enter into some formal arrangement to cooperate with RH and other vendors. I would expect vendors _might_ say, "We don't support OCO modules, and if a problem you ask us to fix is caused by OCO modules there will be an additional fee." What happens if Linux behaves bizarrely and it turns out its misbehavour is caused by a bug in Oracle? I know for sure I can run programs on my toys that result in other programs getting killed off - I did so a couple of days ago, using rsync to backup 20 Gbytes of disk across the LAN. Rsync used more RAM than I think reasonable, and caused other programs to be terminated through lack of virtual memory. It happens that all software involved was standard in the relevant distro, but it needn't have been. -- Cheers John. Join the "Linux Support by Small Businesses" list at http://mail.computerdatasafe.com.au/mailman/listinfo/lssb Copyright John Summerfield. Reproduction prohibited.
