From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jakob Oestergaard Subject: Re: raidreconf feedback help (CFM!) Date: Fri, 8 Nov 2002 00:46:16 +0100 Sender: linux-raid-owner@vger.kernel.org Message-ID: <20021107234616.GC1729@unthought.net> References: <20021106121222.GA23946@vtfl.nu> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Content-Disposition: inline In-Reply-To: <20021106121222.GA23946@vtfl.nu> To: Rikard Morssing Cc: linux-raid@vger.kernel.org List-Id: linux-raid.ids On Wed, Nov 06, 2002 at 01:12:22PM +0100, Rikard Morssing wrote: > I just decided to risk it and expand my raid5-set with raidreconf. It > failed miserably (but that might be due to hw). Ok :) You're not the first to report a failure - and especially on flaky hardware there are things that raidreconf could do to recover gracefully, that it doesn't. >=20 > But it made me think. How should we create meaningful, extensive and > informative debug reports which is crucial for the developers to debu= g > the program? What i think should be on the raidreconf webpage is a li= st > of steps to perform BEFORE and AFTER running the raidreconf, and wher= e > to send the report. "developers" currently means "noone", unfortunately. But I'm probably the closest thing to what you expected it would mean :) I know of a few bugs that can be fixed fairly easily, which would resolve most of the bug reports I have seen. And I am willing to do those fixes. But I have been too busy (for the past year!) to get thes= e things done. It really hurts to see a potentially useful tool rot away like this. The tool did change hands once, to Daniel Cox, who fixed some important bugs that I did not have the time to fix back then. He too, was hit by the "too busy" syndrome, and I got the code back (with his fixes of course), but hasn't touched it much since then. RedHat then took the tool, and did some changes. As far as I know, the only thing they changed was indentation and perhaps some configuration. But I haven't looked more closely, and RedHat never at any point notified me about them using the tool, or even cared to ask if I knew o= f any specific bugs in the tool (go figure...) I tried convincing the EVMS people (who included the kernel RAID functionality in their code as well), to take the raidreconf code and make it do hot-reconfiguration (as part of EVMS). They too, however, seems to be hit by the "too busy" syndrome. (it seems to be contagious - beware!) >=20 > good idea? Yes, very good. Even better; you or anyone else out there listening; if you feel that you have a few weekends to burn, take raidreconf and fix those few bugs that make this tool eat half the arrays that its used on. It really should be no more than a few weekends. I'm sure that there are people on this list who would be more than happ= y to help with testing. And I would of course be delighted to explain the basic architecture of the tool, talk about the few known problems, and come with whatever suggestions I might be able to come up with. If no-one does this - I will fix it myself. I may do so in three weeks from now. Or I may do so in three years from now. I cannot say. I guess there is little doubt, that *if* the tool really worked reliably, it would be useful. But beware. Anyone who's been in contact with the code so far, have bee= n hit by the "too busy" syndrome. I take no responsibility (to the exten= t of applicable law) over any implications contact with the raidreconf code might have on your daily schedule... ;) --=20 =2E............................................................... : jakob@unthought.net : And I see the elder races, : :.........................: putrid forms of man : : Jakob =D8stergaard : See him rise and claim the earth, : : OZ9ABN : his downfall is at hand. : :.........................:............{Konkhra}...............: - To unsubscribe from this list: send the line "unsubscribe linux-raid" i= n the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html