Hello All Me again. Asking about Promise 20276 again. OK, I spent a day or so trying to get the "Promise FastTrak module" to play nice with linux. Unfortunately I haven't succeded YET. Question 1. "Do I need to bother?" I have come to understand that the 2.4.18 kernel has support for this chip and that the 2.4.19 kernel has even better support for this chip. Anyone have experience with this chip and the kernel. I suppose my main concerns are 1. whether or not it is stable 2. how well does it perform. Question 2. How straight forward would it be. I would prepare the RAID arrays first using FastTrak Fastbuild or whatever it's called. Then I would proceed to install SuSE. Correct? On a fresh install could I just pass some boot options and expect to get things working. I have in mind passing info viz the Promise controllers I/O addresses, i.e ide2=0x9400, 0x9002 ide3=... that sort of thing. One last Question. Why not GRUB. All the stuff I have read on setting up raid with linux and the 20276 chip suggest lilo. Any reason for this? tia. t.irvine
irvine@vuosaari.hai.fi wrote:
Hello All
Me again.
Asking about Promise 20276 again.
OK, I spent a day or so trying to get the "Promise FastTrak module" to play nice with linux. Unfortunately I haven't succeded YET.
Question 1. "Do I need to bother?"
I have come to understand that the 2.4.18 kernel has support for this chip and that the 2.4.19 kernel has even better support for this chip.
Anyone have experience with this chip and the kernel. I suppose my main concerns are 1. whether or not it is stable 2. how well does it perform.
I was unable to get it to work under 2.4.18... but it works beautifully under 2.4.19. It's fast and very stable.
Question 2. How straight forward would it be.
I would prepare the RAID arrays first using FastTrak Fastbuild or whatever it's called. Then I would proceed to install SuSE. Correct?
On a fresh install could I just pass some boot options and expect to get things working. I have in mind passing info viz the Promise controllers I/O addresses, i.e ide2=0x9400, 0x9002 ide3=... that sort of thing.
The only thing I added to my config (running SuSE 8.0 with lilo) was: disk = /dev/ataraid/d0 bios = 0x80
One last Question. Why not GRUB. All the stuff I have read on setting up raid with linux and the 20276 chip suggest lilo. Any reason for this?
That's what was installed with 8.0. I've not tried it under 8.1 yet. - Herman
On Fri, Jan 03, 2003 at 10:03:20PM -0800, Herman Knief wrote: Hello
I was unable to get it to work under 2.4.18... but it works beautifully under 2.4.19. It's fast and very stable.
That's good to hear. Thanks for the reply. t.irvine
Hi Irvine. Just wanted to correct something. In 2.4.18 they may list the "FastTrak" modules as being available, and it may be loading, as was the case with me. However, if one looks into the kernel they will see that the model of chipset only goes up to something like xxx60, 62, 72 (something like that). The point being that xxx76 is not listed and hence why it won't fully initialize to a useful state. Also, going to kernel.org before installing 8.1 with kernel version x.x,19 I understood that version x.x18 had not been updated to the present version of the FastTrak but that they had patched the kernel for it in x.x19. Ever since installing the 8.1 with x.x.19 kernel vesion the chipset is fully recognized and inits completely. It behaves correctly and I have not had any issues with it. Furthermore, now that it works properly it initializes before the generic ide interface and one has to use the boot parameter "ide-reverse" so that the proper partition with the kernel and / (root) partition are properly mounted - otherwise you'll get a message that it was unable to find the kernel and booting will stop. With the "ide-reverse" command the boot will look for a second chipset that initializes and then read that - which ironically after getting the pdc20276 chip to work made the generic ide slower and therefore it intialize after the promise chipset make the generic the 2nd - which is why the command is needed. The solution to this situation is to put the HD onto the FastTrak ide connect and install the OS according (i.e as and update) and then it will write to fstab (file system table) correctly. If this were an option for you I would first look at your present fstab and write down all your patition tables. This will change if you mount the HD's to the promise ide's. So something like hda and hdb will become hde and hdf, like wise hdc = hdg, hdd = hdh (everything shifts up by 4). There is a patch somewhere (I can't remember - someone on the list mentioned it) but I'm not very experienced with patching and then reconfigs - I couldn't get it to work. Your best bet is to do as I have done which is install 8.1. Frankly 8.1 is a much better version IMHO and many issues I had with 8.0 have not been a problem in 8.1. In my experience the x.0 versions tend to be the most problematic, such was the case with both 7.0 and 8.0. If you have further questions please write/post them. HTH, Curtis On Friday 03 January 2003 23:54, irvine@vuosaari.hai.fi wrote:
Hello All
Me again.
Asking about Promise 20276 again.
OK, I spent a day or so trying to get the "Promise FastTrak module" to play nice with linux. Unfortunately I haven't succeded YET.
Question 1. "Do I need to bother?"
I have come to understand that the 2.4.18 kernel has support for this chip and that the 2.4.19 kernel has even better support for this chip.
Anyone have experience with this chip and the kernel. I suppose my main concerns are 1. whether or not it is stable 2. how well does it perform.
Question 2. How straight forward would it be.
I would prepare the RAID arrays first using FastTrak Fastbuild or whatever it's called. Then I would proceed to install SuSE. Correct?
On a fresh install could I just pass some boot options and expect to get things working. I have in mind passing info viz the Promise controllers I/O addresses, i.e ide2=0x9400, 0x9002 ide3=... that sort of thing.
One last Question. Why not GRUB. All the stuff I have read on setting up raid with linux and the 20276 chip suggest lilo. Any reason for this?
tia.
t.irvine
-- Billboard Writer vs. Literature = Microsoft vs. Computing,
On Sat, Jan 04, 2003 at 06:07:43AM -0600, Curtis Rey wrote: Hello Curtis Firstly let me say thankyou for your long reply and offer of help. It was much appreciated.
Hi Irvine. Just wanted to correct something. In 2.4.18 they may list the "FastTrak" modules as being available, and it may be loading, as was the case with me. However, if one looks into the kernel they will see that the model of chipset only goes up to something like xxx60, 62, 72 (something like that). The point being that xxx76 is not listed and hence why it won't fully initialize to a useful state.
Good to know. I have SuSE 8.1 so I was planning to use the version of the kernel that came with it.
any issues with it. Furthermore, now that it works properly it initializes before the generic ide interface and one has to use the boot parameter "ide-reverse" so that the proper partition with the kernel and / (root) partition are properly mounted - otherwise you'll get a message that it was unable to find the kernel and booting will stop. With the "ide-reverse" command the boot will look for a second chipset that initializes and then read that - which ironically after getting the pdc20276 chip to work made the generic ide slower and therefore it intialize after the promise chipset make the generic the 2nd - which is why the command is needed.
The system that I have been asked to help with has only two hard disks which we intend to use with the promise controller in a RAID 1 configuration. In others words the intention will be to boot from the RAID array. I found the following howto this morning: http://www.magic-lamp.org/howto/pdc20265-mandrake8.2.phtml Admittedly, it was written for the PDC 20265 chip and for Mandrake users but I read it and feel that the approach suggested in that howto is worth a try. Basically the author recommends 1. installing linux on one of hard disks. This disk should be on a normal ide controller. The 2nd disk should not be connected at all. 2. compiling a new kernel with all the code required to get the RAID array working compiled into the kernel. 3. then making the necessary changes to /etc/lilo and /etc/fstab. 4. halt the computer 5. change the hard disks to the appropriate RAID disk controllers 6. start the machine and make a few changes to the bios - read the link for details 7. continue and setup the RAID arrays with the FastTrack setup utillty 8. "cross your fingers and reboot". His words not mine. Sounds simple enough. I suppose that I will have to think a little about the naming of the drives and how they will change after I reboot, but other than that I have a feeling that this approach may work. Could be wrong, but I will give a try tonight at some stage.
If you have further questions please write/post them.
Thanx, I'll remember that. Thanx again. t.irvine
Is this an onboard interface for the promise chip or is this a pci card? I don't believe believe that you will have to go to such extent to get it to work, especially if it's onboard because the raid arrays will most likely be setup at boot time by the mobo bios - at least it's that way with my Gigabyte GA-7VRXP board. Now the question about how SuSE/Linux will handle this is another story that I can't give any insight into -I have never done this. But I do know that on my board the raid can be setup by the bios and this should make it fairly straight forward for the Linux install/setup. I also believe that Linux will on do raid 1 and not 0, I could be wrong because this may have been overcome and I suspect that the 2.6.x kernel may fix this is it isn't already. I know that my onboard chip only does these two types. If your going for mirror (raid 1) then it should be simple enough (famous last words). It should be in the mobo manual (mine is). Now as for a pci card I don't know. I have no knowledge of getting a pci card to boot and how to set the arrays up - probably need to install a software program to do this which of course will undoubtedly be Windblow base. Have a look at this site: http://www.murty.net/ataraid/ I couldn't find alot in goolge YMMV. HTH, Curtis On Saturday 04 January 2003 11:08, irvine@vuosaari.hai.fi wrote:
On Sat, Jan 04, 2003 at 06:07:43AM -0600, Curtis Rey wrote: Hello Curtis
Firstly let me say thankyou for your long reply and offer of help. It was much appreciated.
Hi Irvine. Just wanted to correct something. In 2.4.18 they may list the "FastTrak" modules as being available, and it may be loading, as was the case with me. However, if one looks into the kernel they will see that the model of chipset only goes up to something like xxx60, 62, 72 (something like that). The point being that xxx76 is not listed and hence why it won't fully initialize to a useful state.
Good to know. I have SuSE 8.1 so I was planning to use the version of the kernel that came with it.
any issues with it. Furthermore, now that it works properly it initializes before the generic ide interface and one has to use the boot parameter "ide-reverse" so that the proper partition with the kernel and / (root) partition are properly mounted - otherwise you'll get a message that it was unable to find the kernel and booting will stop. With the "ide-reverse" command the boot will look for a second chipset that initializes and then read that - which ironically after getting the pdc20276 chip to work made the generic ide slower and therefore it intialize after the promise chipset make the generic the 2nd - which is why the command is needed.
The system that I have been asked to help with has only two hard disks which we intend to use with the promise controller in a RAID 1 configuration. In others words the intention will be to boot from the RAID array.
I found the following howto this morning:
http://www.magic-lamp.org/howto/pdc20265-mandrake8.2.phtml
Admittedly, it was written for the PDC 20265 chip and for Mandrake users but I read it and feel that the approach suggested in that howto is worth a try.
Basically the author recommends 1. installing linux on one of hard disks. This disk should be on a normal ide controller. The 2nd disk should not be connected at all.
2. compiling a new kernel with all the code required to get the RAID array working compiled into the kernel.
3. then making the necessary changes to /etc/lilo and /etc/fstab.
4. halt the computer
5. change the hard disks to the appropriate RAID disk controllers
6. start the machine and make a few changes to the bios - read the link for details
7. continue and setup the RAID arrays with the FastTrack setup utillty
8. "cross your fingers and reboot". His words not mine.
Sounds simple enough. I suppose that I will have to think a little about the naming of the drives and how they will change after I reboot, but other than that I have a feeling that this approach may work. Could be wrong, but I will give a try tonight at some stage.
If you have further questions please write/post them.
Thanx, I'll remember that.
Thanx again.
t.irvine
-- Billboard Writer vs. Literature = Microsoft vs. Computing,
On Sat, Jan 04, 2003 at 02:21:48PM -0600, Curtis Rey wrote:
Is this an onboard interface for the promise chip or is this a pci card?
Onboard.
I don't believe believe that you will have to go to such extent to get it to work, especially if it's onboard because the raid arrays will most likely be setup at boot time by the mobo bios - at least it's that way with my Gigabyte GA-7VRXP board. Now the question about how SuSE/Linux will handle this is another story that I can't give any insight into -I have never done this. But I do know that on my board the raid can be setup by the bios and this should make it fairly straight forward for the Linux install/setup. I also believe that Linux will on do raid 1 and not 0, I could be wrong because this may have been overcome and I suspect that the 2.6.x kernel may fix this is it isn't already. I know that my onboard chip only does these two types. If your going for mirror (raid 1) then it should be simple enough (famous last words). It should be in the mobo manual (mine is).
You could be right about this method not being the quickest. In my first attempt to get this card working I started installing with both drives connected to the onboard promise connector interfaces. After playing with the bios 4 a long time I got the thing to boot and all seemed ok, but I got a little confused about what to do next. Then my friend came along and we decided to try another approach. You can see the email I wrote at the following: http://lists.suse.com/archive/suse-linux-e/2003-Jan/0277.html
Have a look at this site:
Thanks for the link. I take a look.
I couldn't find alot in goolge YMMV.
And a big thanx for taking the time to look. Take care. t.irvine
irvine@vuosaari.hai.fi wrote:
On Sat, Jan 04, 2003 at 02:21:48PM -0600, Curtis Rey wrote:
Is this an onboard interface for the promise chip or is this a pci card?
Onboard.
I don't believe believe that you will have to go to such extent to get it to work, especially if it's onboard because the raid arrays will most likely be setup at boot time by the mobo bios - at least it's that way with my Gigabyte GA-7VRXP board. Now the question about how SuSE/Linux will handle this is another story that I can't give any insight into -I have never done this. But I do know that on my board the raid can be setup by the bios and this should make it fairly straight forward for the Linux install/setup. I also believe that Linux will on do raid 1 and not 0, I could be wrong because this may have been overcome and I suspect that the 2.6.x kernel may fix this is it isn't already. I know that my onboard chip only does these two types. If your going for mirror (raid 1) then it should be simple enough (famous last words). It should be in the mobo manual (mine is).
You could be right about this method not being the quickest. In my first attempt to get this card working I started installing with both drives connected to the onboard promise connector interfaces.
After playing with the bios 4 a long time I got the thing to boot and all seemed ok, but I got a little confused about what to do next. Then my friend came along and we decided to try another approach. You can see the email I wrote at the following:
http://lists.suse.com/archive/suse-linux-e/2003-Jan/0277.html
Have a look at this site:
Thanks for the link. I take a look.
I couldn't find alot in goolge YMMV.
And a big thanx for taking the time to look.
Take care.
t.irvine
Any time. Hope it's working, Cheers, Curtis. P.S. post agian if you get it working or not so it can be archived for future references. :)
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Sunday 05 January 2003 08:17, Curtis Rey wrote:
P.S. post agian if you get it working or not so it can be archived for future references. :)
a friend of me is busy with the dutch-to-english translation of the howto from my site, if he is ready, i will post it here. - -- greetz Frederik Vos L4L & VosBerg@SuSE http://www.vosberg.be -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.0.7 (GNU/Linux) iD8DBQE+F/LZRCeyhsvrMycRAjg1AJ4wOnoldN8kByQPj7V4fP92UZ6uEQCgpc0v T9YioRX7B4vJJxlq5VotLtI= =CEUP -----END PGP SIGNATURE-----
participants (4)
-
Curtis Rey -
Frederik Vos -
Herman Knief -
irvine＠vuosaari.hai.fi