Re: Status of Debian on Sun Blade 150
On Tue, 26 Apr 2005, Jurij Smakov wrote: Hi Wiktor, [snip] According to the logs the base-installer (it's postinst script, actually) exited with error cannot_install. According to the source, this error can only occur (given that the CD is properly mounted under /cdrom) if it fails to read the contents of the /cdrom/.disk/base_components file (which is present on CD). I tend to think that it was just some random cd-rom glitch (I didn't encounter any problems during my testing). Could you please try reproducing it, and if the problem persists, return to main menu, start a shell and check that the mentioned file indeed exists and contains the string 'main' (without quotes)? Now I didn't really expect this, but after booting a machine with the netboot image you provided and switching to the 2nd console I got: # ls -a /cdrom/.disk . .. base_components base_installableinfoudeb_include # cat /cdrom/.disk/base_components cat: Read error: Input/output error You are pure genius! I'll try another burning session soon. And maybe another disk. At least I know what to look for. BTW. I've found a similar behaviour on another testing machine. It concerns the partitioner. I let it to format partitions, but later I can't get it to display properly the contents of the drive. I prefer doing the manual partitioning, but partitions don't show up. Weird. It happened twice on two different machines, but not once on the third one. Do you think it may be related to te same CD problem? Thanks anyway! Wiktor Wandachowicz -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Status of Debian on Sun Blade 150
Frans Pop wrote: Hi Wiktor, On Wednesday 27 April 2005 19:39, Wiktor Wandachowicz wrote: And the problem I described with the parted is consistent on the same machine. I can't get it to display partitions after I select manual partitioning from its menu. I'm afraid this is user error. You've put swap as the first partition (starting on cylinder 0). This overwrites the sun-disklabel that is written in that location! The result is that parted no longer recognizes the disk during the second installation. With sun-disklabel you are not allowed to put swap or RAID or LVM at the start of the disk. ext2/ext3 are OK. See also: http://d-i.alioth.debian.org/manual/en.sparc/apbs05.html#id2538449 I didn't do it. Solaris 9 installer did it! It is the one who put swap space at clusters 0-258: # fdisk -l /dev/hda Disk /dev/hda (Sun disk label): 16 heads, 255 sectors, 19156 cylinders Units = cylinders of 4080 * 512 bytes Device FlagStart EndBlocks Id System /dev/hda1 258 319124440 83 Linux native /boot /dev/hda2 u 0 2585263203 SunOS swap /dev/hda3 0 19156 390782405 Whole disk /dev/hda4 15391 19156 76806002 SunOS root /dev/hda5 3660 6744 6291360 83 Linux native / (sarge1) /dev/hda6 6744 10574 7813200 83 Linux native /home /dev/hda7 10574 15391 9826680 83 Linux native / (sarge2) /dev/hda8 319 3660 68156408 SunOS home BTW, that machine ALREADY WORKED before trying netboot. It had Solaris installed first (hda{2,4,8}), with manually specified partition sizes, then Debian has been installed side-by-side with Solaris (hda{1,5,6}). All partitions have been visible before trying netboot, and they are still visible after installing second Debian on spare (hda7) partition. And fdisk still works too, just as it worked before testing the installation. The machine still boots, too, both into sarge1 as well as sarge2. (I can't put silo.conf here, because all machines are turned off ATM) User error. That's impolite. 'Cause I've spent LOTS of time crafting that multiboot configuration. The best proof is that installer DID everything correctly first time (when all partitions were right), then after formatting swap (hda2) it turned out that it was unable to display partitions list anymore. From my POV it is the one to blame. If it is indeed user error (I mean: the Solaris installer's and my error), then I'm terribly sorry. But in that case please elaborate more how it should be done The Right Way. Keep in mind that multiboot with a spare partition is a must. P.S. If you want fdisk, just switch to VT2, wget the fdisk udeb from a mirror [1] and 'udpkg -i' it. [1] http://ftp.debian.org/debian/pool/main/u/util-linux/ Wow, I didn't know that. Thank you very much! With best regards, Wiktor Wandachowicz -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Status of Debian on Sun Blade 150
On Thu, 28 Apr 2005, Wiktor Wandachowicz wrote: Frans Pop wrote: With sun-disklabel you are not allowed to put swap or RAID or LVM at the start of the disk. ext2/ext3 are OK. Frans, if you are referring to #283303, I thought that we've fixed it. At least the bug is marked 'fixed' when version 62 of partman has been uploaded, and I remember testing it and confirming that the bug is gone. Version in RC3 is 63, so it should be fixed there too. I didn't do it. Solaris 9 installer did it! It is the one who put swap space at clusters 0-258: # fdisk -l /dev/hda Disk /dev/hda (Sun disk label): 16 heads, 255 sectors, 19156 cylinders Units = cylinders of 4080 * 512 bytes Device FlagStart EndBlocks Id System /dev/hda1 258 319124440 83 Linux native /boot /dev/hda2 u 0 2585263203 SunOS swap /dev/hda3 0 19156 390782405 Whole disk /dev/hda4 15391 19156 76806002 SunOS root /dev/hda5 3660 6744 6291360 83 Linux native / (sarge1) /dev/hda6 6744 10574 7813200 83 Linux native /home /dev/hda7 10574 15391 9826680 83 Linux native / (sarge2) /dev/hda8 319 3660 68156408 SunOS home Wow, this is one complicated partition configuration. I would not be too surprised that partman chokes on it. I was trying to suggest to create a clean disklabel before partitioning, but as I understand it is out of the question =). User error. That's impolite. 'Cause I've spent LOTS of time crafting that multiboot configuration. The best proof is that installer DID everything correctly first time (when all partitions were right), then after formatting swap (hda2) it turned out that it was unable to display partitions list anymore. From my POV it is the one to blame. If it is indeed user error (I mean: the Solaris installer's and my error), then I'm terribly sorry. But in that case please elaborate more how it should be done The Right Way. Keep in mind that multiboot with a spare partition is a must. Everyone is working towards the same goal here, so let's not get too picky about the choice of words. Please try partitioning with fdisk, as Frans suggested. If it will work as expected, we can be pretty confident that the problem lies with partman. Also, it would be interesting to see whether parted and fdisk recognize the partition sizes and types correctly on a runnning system. Thanks and best regards, Jurij Smakov[EMAIL PROTECTED] Key: http://www.wooyd.org/pgpkey/ KeyID: C99E03CC -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: disk size mismatch issue during install on ultra1
Thanks for all the info. I zorched the partition table and everything went smooth from there. I also cleaned the other 9GB disk in the u1 and both fdisk and parted report things as they should be. Thanks, Ardo [EMAIL PROTECTED] ([EMAIL PROTECTED]) wrote: One trick to consider. If it's not a boot disk, use an MSDOS / PC label. The kernel can still read it and it's a lot less flaky than the Sun scheme. One issue I have had with large disks and Sun labels (but this only occurred on a 40G volume) is that the disk did not work right with a Sun label, and I had to use PC instead to get it to work at all. Can another helpful D-I or Sun internals person talk a little about disk labels? They seem to be a little odd and I do not understand why they cause such problems. Another thing to try: zorch the hell out of the disk label with dd if=/dev/zero of=/dev/scsi/host0/bus0/target0/lun0/disc bs=1M Then redo it with parted's CHS numbers: 8637,64,32. I have noticed parted and SPARC fdisk tend to take the label as gospel truth even when it's stuffed full of inaccurate crap. I bet if you erase the label and recreate it, this problem should stop happening. That was another trick I needed to use to get going at one point. Another thing, get the proper CHS numbers from the disk maker and type those in to the label creation command in parted. This should help you with your geometry problem. HTH! --- Ardo van Rangelrooij [EMAIL PROTECTED] wrote: [EMAIL PROTECTED] ([EMAIL PROTECTED]) wrote: I had an issue similar to this one when installing on an Ultra 2 with a similar setup. I found I could get around the problem by selecting the execute a shell option in the debian installer, then running parted on the disk manually. One thing I cannot remember that you will need to find out in order to do this is the proper device name for your disks. The name convention used by the installer kernel is really bizarre and I do not understand how it works. Hopefully the D-I people on this list will be able to explain what device names your disks might be using, or you can reverse engineer it from the install screens. --- Ardo van Rangelrooij [EMAIL PROTECTED] wrote: Hi, I've got an interesting issue with installing sarge on an ultra1: At the beginning of the partitioning the 9GB HD's are detected correctly as having 9GB. However, when configuring the 1st partition as 4GB, the partition size is reported as 2GB in the partition info screen. When I continue there's only some 27MB left. When using guided partitioning I get similar results (1.9GB + 99MB) and the partitioning even fails with an error message stating that there are too many primary partitions. I'm using the netboot from April 24 and both 2.4 and 2.6 fail. Older netboots (e.g. rc3) also fail here. The ultra1 has OBP v3.35 and probe-scsi detects the HD's correctly. It has 256MB memory. Any thoughts/hints/help how to get across this hurdle? I looked through the bug reports for d-i but didn't find anything about this issue. Thanks for the tip. I'm a step closer. When running parted by hand I get /bin # parted /dev/scsi/host0/bus0/target0/lun0/disc print Warning: The disk CHS geometry (8637,64,32) does not match the geometry stored on the disk label (1107,27,133). Ignore/Cancel? c Information: Don't forget to update /etc/fstab, if necessary. Unfortunately running parted /dev/scsi/host0/bus0/target0/lun0/disc mklabel sun does NOT resolve this. The geometry mismatch persits. I'll run fdisk on this disk in another machine and see what gives. Thanks, Ardo -- Ardo van Rangelrooij Debian XML/SGML Group [EMAIL PROTECTED] [EMAIL PROTECTED] http://people.debian.org/~ardo/ http://debian-xml-sgml.alioth.debian.org/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED] -- Ardo van Rangelrooij Debian XML/SGML Group [EMAIL PROTECTED] [EMAIL PROTECTED] http://people.debian.org/~ardo/ http://debian-xml-sgml.alioth.debian.org/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: WG: Re: Bits (Nybbles?) from the Vancouver release team meeting
Linux qux 2.6.11.5 #1 Wed Mar 23 13:23:49 PST 2005 sparc64 GNU/Linux --- David S. Miller [EMAIL PROTECTED] wrote: On Tue, 26 Apr 2005 23:01:57 -0700 (PDT) [EMAIL PROTECTED] wrote: The dmesg outputs motivating my original (misinterpreted and/or poorly phrased) IDE controller rant: hdc: dma_timer_expiry: dma status == 0x60 hdc: DMA timeout retry hdc: timeout waiting for DMA hdc: status error: status=0x58 { DriveReady SeekComplete DataRequest } hdc: drive not ready for command What kernel version gives you this? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]