On 3/5/08, James Bottomley <James.Bottomley@xxxxxxxxxxxxxxxxxxxxx> wrote: > On Wed, 2008-03-05 at 13:52 -0300, Rodrigo Severo wrote: > > On 3/5/08, James Bottomley <James.Bottomley@xxxxxxxxxxxxxxxxxxxxx> wrote: > > > On Wed, 2008-03-05 at 13:26 -0300, Rodrigo Severo wrote: > > > > On 3/5/08, Salyzyn, Mark <Mark_Salyzyn@xxxxxxxxxxx> wrote: > > > > > Sounds like you have a bad SCSI cable or drive. > > > > > > > > > > Do any of these drives work in any other OS environment? > > > > > > > > This same drive with this same cable with this same terminator works > > > > well with a 39160 (which usses another driver: the aic7xxx) on the > > > > same OS environment so I don't think this is a cable or drive issue. > > > > > > That's only ultra 160 capable. What happens if you take the speeds for > > > the drives down in the bios of the 29320? > > > > I changed the following values on 29320LPE from the previous default ones: > > > > Sync Transfer Rate: 160 > > Packetized: No > > QAS: No > > > > The error messages are unchanged. > > That does point further to card problems. > > > > > Besides I tested with four different cables (different models and > > > > manufactures, all pretend to be U320 compatible). The problem is > > > > unchanged when I use different cables and terminators. > > > > > > > > I also tested with several different disks (different models and > > > > manufactures). The problem is still unchanged. > > > > > > It's sounding a bit more like a damaged board ... but try reducing the > > > speed in the bios. > > > > As I said before, you have much more experience on SCSI issues but I > > still can't see how a damaged SCSI board would work on one motherboard > > model. Do you really think this is possible? > > > Certainly ... depends on the problem ... for instance, you could have a > damaged trace in the high 32 bits and not see the problem when you put > it in a 32 bit slot. Likewise, it could be a capacitance problem on the > traces and your motherboard that works has a lower PCI bus speed. I have not mentioned that this board - the 29320LPE - is a 1x PCI Express, not a PCI-X, please forgive me. Back to the issue at hand, being a 1x PCI Express board there can be no lines used when installed on one motherboard and not used when on another motherboard AFAICT. Also I am not aware of any different bus speeds for the PCI Express bus I'm not a PCI Express specialist so I may be wrong here. > The fact that it works in one system tends to rule out software driver > problems. If you've tried different cables and reducing the speed, that > tends to eliminate the interconnect, so all that's left is the card (and > the motherboards, I suppose, although three failing would tend to rule > them out). I understand your line of reasoning. But I think this is not the case as I gathered several reports on the net from users with this same SCSI board having problems with several chipsets and managing to make it work on some few specific motherboards. This is what's happening to me. As far as I can tell none managed to get a usefull statement from Adaptec on the subject so I really can't tell if this is a SCSI board problem, a chipset problem or even a driver problem. Several tried. I did. Got absolutely no answer at all from Adaptec, only dead silence. Anyway, I might have a broken SCSI board but the same exactly fact that convinces you that the board is malfunctiong (work in one motherboard and not in three others) is the kind of argument that I'm sure Adaptec will use to argue the opposite: the SCSI board must be working well if it works in one motherboard. Here are some tests I already did: On the motherboard that the PCI-e 29320LPE card worked I managed to copy several (>70) giga of data copied onto disks connected to it and then start the MySQL server that used this data. It worked for less than a hour before I got a kernel panic. At this point I just changed the SCSI board to a PCI-X 39160. I got another kernel panic in less than a hour. I was using plain kernel 2.6.24.3. I got back to a Gentoo specific kernel 2.6.22 and the same 39160 board is working perfect for more than 48 hours. The PCI-e 29320LPE panics after 15 minutes or so with this last kernel. I didn't took note of the details of the kernel panics but they looked quite the same to me: both the PCI-e 29320LPE and the PCI-X 39160. This makes me believe there are other things wrong. I would like to iron these out before I decide the new PCI-e 29320LPE is broken. I'm just not sure how to do it. Rodrigo Severo -- --------------------------------------------------------------------------------------- Rodrigo Severo Fábrica de Idéias SBS Quadra 2 - Bloco S - Ed. Empire Center - Sala 1.301 Brasília - DF - CEP 70070-904 Tel. (61) 3321-1357 Fax (61) 3223-1712 --------------------------------------------------------------------------------------- -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
