Re: ECKD driver vs DIAG driver
On 04.01.2011 16:02, Richard Troth wrote: Mark is correct: one automagically created partition. Worse, there is no 'fdasd' or 'fdisk' management of that partition. Actually you *can* use fdisk to manage partitions on DASDs which are formatted with a fixed block size (ECKD LDL or FBA), regardless of whether DIAG access is used or not. Example: to get rid of the implicitly created DASD partitions on /dev/dasdx, use these commands: echo w | fdisk /dev/dasdx This will write an MS-DOS type master boot record with an empty partition table to dasdx. Note that some operations (mkfs.ext, pvcreate) seem to overwrite the MBR so the fdisk step should be performed last. More background: When a DASD is set online, the Linux kernel attempts to recognize the partition type by letting all supported partition type handlers have a look at the disk contents. This process happens sequentially and stops once a handler indicates that it has found a valid partition. The DASD partition handler comes very late in the list, so pretty much any partitioning scheme supported by the Linux kernel can be used to circumvent implicit DASD partitioning. WORSE STILL, you *must* put the filesystem into the partition (such as it is) if you are going to boot from this disk. A filesystem in /dev/dasdx will be clobbered by the first stage of the boot loader, while a filesystem in /dev/dasdx1 is protected by the extra 8K of padding. (12K total) I checked it again this morning. The bootstrap overwrites the root inode. This is a side effect of the zipl boot loader design, more precisely of the size of the first stage IPL code that is written to block 0. It's conceivable that this could be changed in the future. God bless whoever in Boeblingen fixed this problem for FBA disks. You can use the pseudo-partition, or not. You can boot from them either way. This problem never existed for FBA disks - the first stage FBA IPL code only spans a single FBA block. According to Martin Schwidefsky this approach was chosen because there were no format or layout restrictions for FBA disk so they decided to keep it similar to the Intel world. Regards, Peter Oberparleiter -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: ECKD driver vs DIAG driver
Actually you *can* use fdisk to manage partitions on DASDs ... Example: to get rid of the implicitly created DASD partitions on /dev/dasdx, use these commands: echo w | fdisk /dev/dasdx Nice. THANK YOU. A quick check confirms that it works for both EXT2 and ISO-9660 filesystems. You have to create the filesystem first, of course. For example, 'mke2fs' will write zeros in the first 1K. (ISO FS starts at 32K, I believe.) But 'fdisk' safely keeps its PC partition table down to 512 bytes, so it does not clobber the FS. Do the 'fdisk' after the 'mke2fs' and it works like a charm. -- R; Rick Troth Velocity Software http://www.velocitysoftware.com/ On Mon, Jan 10, 2011 at 08:25, Peter Oberparleiter ober...@linux.vnet.ibm.com wrote: On 04.01.2011 16:02, Richard Troth wrote: Mark is correct: one automagically created partition. Worse, there is no 'fdasd' or 'fdisk' management of that partition. Actually you *can* use fdisk to manage partitions on DASDs which are formatted with a fixed block size (ECKD LDL or FBA), regardless of whether DIAG access is used or not. Example: to get rid of the implicitly created DASD partitions on /dev/dasdx, use these commands: echo w | fdisk /dev/dasdx This will write an MS-DOS type master boot record with an empty partition table to dasdx. Note that some operations (mkfs.ext, pvcreate) seem to overwrite the MBR so the fdisk step should be performed last. More background: When a DASD is set online, the Linux kernel attempts to recognize the partition type by letting all supported partition type handlers have a look at the disk contents. This process happens sequentially and stops once a handler indicates that it has found a valid partition. The DASD partition handler comes very late in the list, so pretty much any partitioning scheme supported by the Linux kernel can be used to circumvent implicit DASD partitioning. WORSE STILL, you *must* put the filesystem into the partition (such as it is) if you are going to boot from this disk. A filesystem in /dev/dasdx will be clobbered by the first stage of the boot loader, while a filesystem in /dev/dasdx1 is protected by the extra 8K of padding. (12K total) I checked it again this morning. The bootstrap overwrites the root inode. This is a side effect of the zipl boot loader design, more precisely of the size of the first stage IPL code that is written to block 0. It's conceivable that this could be changed in the future. God bless whoever in Boeblingen fixed this problem for FBA disks. You can use the pseudo-partition, or not. You can boot from them either way. This problem never existed for FBA disks - the first stage FBA IPL code only spans a single FBA block. According to Martin Schwidefsky this approach was chosen because there were no format or layout restrictions for FBA disk so they decided to keep it similar to the Intel world. Regards, Peter Oberparleiter -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/ -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: ECKD driver vs DIAG driver
The Linux kernel does not consider an LDL-format disk to be unpartitioned. But the partition you see is a phantom. And you can use an LDL-format disk as if it were unpartitioned. We're battling semantics, or something along those lines. I previously used the term partition zero. People did not seem to understand that. The whole disk seems to work. It's good to be able to use the whole disk, if only for reduced complexity. It's bad when the bootstrap clobbers the filesystem. Here's a rough diagram: + - + CDL + LDL + FBA + | filesystem in partition 0 | --- | -Y- | -Y- | | bootable with fs in part0 | --- | --- | -Y- | | filesystem in partition 1 | -Y- | -Y- | -Y- | | bootable with fs in part1 | -Y- | -Y- | -Y- | | filesystem in partition 2 | -Y- | --- | --- | | bootable with fs in part2 | -Y- | --- | --- | | filesystem in partition 3 | -Y- | --- | --- | | bootable with fs in part3 | -Y- | --- | --- | + - + --- + --- + --- + | works with DIAG250 driver | --- | -Y- | -Y- | + - + --- + --- + --- + | --- external transparency | --- | --- | -Y- | + - + --- + --- + --- + (Looks really bad with my proportional font email interface. See below for a Googoo doc.) I advocate use of partition zero for filesystems. It is a little-known and underutilized feature of zLinux just like the CMS RESERVEd file. Put a filesystem on /dev/dasdq, reboot, and you'll still see a /dev/dasdq1 partition. But the filesystem survives in /dev/dasdq. This is a Good Thing. In this case, just ignore /dev/dasdq1. It is an artifact of the driver. So ... to answer the question about what Rick is asking for, I want the bootable with fs in part0 to change to Y for LDL. It requires a ZIPL change. Another little-known and underutilized fact is the internal/external transparency of FBA storage. (Talk about reduced complexity!) I will spare the group and not discuss it further except to say that it is ... just another: Something that doesn't get advertised enough, kind of like CMS RESERVE. I have attempted to collect some of this info into a spread sheet: http://spreadsheets.google.com/ccc?key=0ArAkQhEvbQZfdEhxZDhLNEEwU0dvVlJBUmVXVnJ6c1E Not sure how to fit use of CMS RESERVE on that. Suggestions? -- R; Rick Troth Velocity Software http://www.velocitysoftware.com/ On Thu, Jan 6, 2011 at 16:47, Stephen Powell zlinux...@wowway.com wrote: On Thu, 06 Jan 2011 16:10:36 -0500 (EST), Richard Troth wrote: The problem is that one cannot boot from an unpartitioned CKD disk (LDL) even though one can boot from an unpartitioned FBA disk. Partition tables are not required for other disks and bootstraps. Why should they be required for mainframe disks and bootstraps? The Linux kernel does not consider an LDL-format disk to be unpartitioned. If you format a disk with dasdfmt using -d ldl (and other appropriate parameters), then the disk has been implicitly partitioned, as far as the Linux kernel is concerned. Assuming CKD DASD, the implicit partition will begin with the fourth physical block. (The first two blocks are reserved for IPL records, the third block is the volume label.) I haven't tested your exact scenario, but here's what I have tested. I have a Linux machine that runs in a virtual machine under z/VM. It has four disks, as follows: device block mount number special point file -- --- - 0200 /dev/dasda /dev/dasda1 / 0201 /dev/dasdb /dev/dasdb1 /boot 0202 /dev/dasdc /dev/dasdc1 /home 0203 /dev/dasdd /dev/dasdd1 swap All four of the disks are CMS reserved minidisks. All of them use the DIAG driver except 0201, which uses the ECKD driver. The boot device is 0201. Linux is started by IPL 0201 It works great. I've been doing it for years. What's the problem? (0201 has to use the ECKD driver because zipl does not support writing IPL records to a device controlled by the DIAG driver) -- .''`. Stephen Powell : :' : `. `'` `- -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/ -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: ECKD driver vs DIAG driver
On Fri, 07 Jan 2011 20:06:25 -0500 (EST), Richard Troth wrote: Stephen Powell wrote: ... The Linux kernel does not consider an LDL-format disk to be unpartitioned. ... But the partition you see is a phantom. And you can use an LDL-format disk as if it were unpartitioned. We're battling semantics, or something along those lines. I previously used the term partition zero. People did not seem to understand that. The whole disk seems to work. OK, now I think I understand what you are talking about. You're swimming upstream, Rick. That's not the way things were designed to work. Linux grew up in the Intel 386 world, where hard disks were always assumed to be partitioned. IBM mainframe DASD did not fit that pattern; so a way was invented to partition mainframe DASD. For cdl, an OS-style VTOC acted as the partition table and a dataset acted as a partition. (Of course, some restrictions were added. The dataset had to consist of a single extent, and a maximum of three datasets could be defined.) For LDL and CMS formats, there is no partition table. The single partition on the disk is implicit, rather than explicit. Nevertheless, it is there. You should always make your file systems (mkfs et al) or swap spaces (mkswap) on a PARTITION; never on a DEVICE. The only thing in your system that should refer to a DEVICE is zipl, when it writes out IPL records. IPL records are written by zipl to a DEVICE, not a PARTITION. The concept of a partition boot sector from the i386 world was really never carried over into the mainframe world. Mainframe DASD partitions don't have boot sectors. In the mainframe world, only the master boot record (the device itself) can be IPLed. I understand what you want to do, but that is not how things were designed to work. You're trying to force a square peg into a round hole. If you follow the advice of the above paragraph, you should never have any problems. -- .''`. Stephen Powell : :' : `. `'` `- -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: ECKD driver vs DIAG driver
In theory, we could mangle the IPL code into the first 512 bytes of a filesystem which is reserved on x86 as partition boot record, ... You should thusly mangle the IPL code. :-) It doesn't break anything else to do so. a) the disk would not be labled as observed by other mainframe OSes b) this would differ from how people use Linux on other platforms where they typically do use partitions ... The problem with mainframe labelling (the req for an IBM volser) is that it panders to the bad behaviour of a certain other mainframe op sys. Also, we're mixing labelling with partitioning. They are not the same thing, though CDL accommodates both. It is presumptuous, and increasingly incorrect, to expect that users partition everything on other platforms. In fact, there are growing numbers of cases where users/customers DO NOT partition physical media. Windows assumes partitioning (except where it is forced to know better). Too bad for Windows. z/OS assumes IBM volsers. Too bad for z/OS. This is Linux. We should be able to bypass the complexities and ... Keep It Simple. Partitioning is a layer of complexity. Where it is needed, fine. Where it is not needed, let us turn it off. -- R; Rick Troth Velocity Software http://www.velocitysoftware.com/ On Wed, Jan 5, 2011 at 03:06, Carsten Otte co...@de.ibm.com wrote: Quote: Mark is correct: one automagically created partition. Worse, there is no 'fdasd' or 'fdisk' management of that partition. WORSE STILL, you *must* put the filesystem into the partition (such as it is) if you are going to boot from this disk. A filesystem in /dev/dasdx will be clobbered by the first stage of the boot loader, while a filesystem in /dev/dasdx1 is protected by the extra 8K of padding. (12K total) I checked it again this morning. The bootstrap overwrites the root inode. Well the LDL disk layout basically consists of two blocks of data (size depends on the blocksize used for formatting) Those are being used to a) label the device and b) conatin the IPL boot code (channel programs). In theory, we could mangle the IPL code into the first 512 bytes of a filesystem which is reserved on x86 as partition boot record, but that'd have a couple of downsides: a) the disk would not be labled as observed by other mainframe OSes b) this would differ from how people use Linux on other platforms where they typically do use partitions in front of the partition. Note that for LDL formatted media, you may chose to put the filesystem on the device itself (like /dev/dasdx instead of /dev/dasdx1) for volumes that you do not intend to boot from. Due to the fact that for ECKD CDL media blocks on track 0 do not have the formatted size, you cannot do the same with ECKD CDL formatted disks (filesystem corruption would be the result). with kind regards Carsten Otte IBM Linux Technology Center / Boeblingen lab -- omnis enim res, quae dando non deficit, dum habetur et non datur, nondum habetur, quomodo habenda est -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/ -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: ECKD driver vs DIAG driver
On Thu, 06 Jan 2011 09:03:46 -0500 (EST), Richard Troth wrote: ... The problem with mainframe labelling (the req for an IBM volser) is that it panders to the bad behaviour of a certain other mainframe op sys. Also, we're mixing labelling with partitioning. They are not the same thing, though CDL accommodates both. It is presumptuous, and increasingly incorrect, to expect that users partition everything on other platforms. In fact, there are growing numbers of cases where users/customers DO NOT partition physical media. Windows assumes partitioning (except where it is forced to know better). Too bad for Windows. z/OS assumes IBM volsers. Too bad for z/OS. This is Linux. We should be able to bypass the complexities and ... Keep It Simple. Partitioning is a layer of complexity. Where it is needed, fine. Where it is not needed, let us turn it off. Hello, Rick. I have seen this thread running for some time now, and I'm getting more and more confused. What, exactly, is the problem that you are trying to solve? I have recently completed work on enhancements to GNU parted to support all combinations of DASD type (CKD and FBA) DASD format (cdl, ldl, CMS non-reserved, and CMS reserved) and DASD driver (ECKD, FBA, and DIAG) supported by the Linux kernel; so I am reasonably familiar with the landscape here. I have not had any problems. Can you please state your problem in a manner simple enough for a dunder head like me to understand? -- .''`. Stephen Powell : :' : `. `'` `- -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: ECKD driver vs DIAG driver
On 1/6/2011 at 02:22 PM, Stephen Powell zlinux...@wowway.com wrote: I have recently completed work on enhancements to GNU parted to support all combinations of DASD type (CKD and FBA) DASD format (cdl, ldl, CMS non-reserved, and CMS reserved) and DASD driver (ECKD, FBA, and DIAG) supported by the Linux kernel; Where do you live? I'd like to buy you a drink. :) Mark Post -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: ECKD driver vs DIAG driver
The problem is that one cannot boot from an unpartitioned CKD disk (LDL) even though one can boot from an unpartitioned FBA disk. Partition tables are not required for other disks and bootstraps. Why should they be required for mainframe disks and bootstraps? Anyway ... I am hopeful that Boeblingen can fix it faster than any of us would-be contributors because they invented 'zipl' in the first place. Also, they already fixed it to work with unpartitioned FBA, so someone took a big step in the right direction. CDL is related, but is a different ... er, uh ... problem. Fixing the bootstrap ('zipl') won't fix that. THANK YOU for your work on 'parted'. Having full S/390 support there will be terrific. Maybe I can now talk you into contributing to GRUB in the other direction. (ie: so that it will work with unpartitioned disks like other PC boostraps do) -- R; Rick Troth Velocity Software http://www.velocitysoftware.com/ On Thu, Jan 6, 2011 at 14:22, Stephen Powell zlinux...@wowway.com wrote: On Thu, 06 Jan 2011 09:03:46 -0500 (EST), Richard Troth wrote: ... The problem with mainframe labelling (the req for an IBM volser) is that it panders to the bad behaviour of a certain other mainframe op sys. Also, we're mixing labelling with partitioning. They are not the same thing, though CDL accommodates both. It is presumptuous, and increasingly incorrect, to expect that users partition everything on other platforms. In fact, there are growing numbers of cases where users/customers DO NOT partition physical media. Windows assumes partitioning (except where it is forced to know better). Too bad for Windows. z/OS assumes IBM volsers. Too bad for z/OS. This is Linux. We should be able to bypass the complexities and ... Keep It Simple. Partitioning is a layer of complexity. Where it is needed, fine. Where it is not needed, let us turn it off. Hello, Rick. I have seen this thread running for some time now, and I'm getting more and more confused. What, exactly, is the problem that you are trying to solve? I have recently completed work on enhancements to GNU parted to support all combinations of DASD type (CKD and FBA) DASD format (cdl, ldl, CMS non-reserved, and CMS reserved) and DASD driver (ECKD, FBA, and DIAG) supported by the Linux kernel; so I am reasonably familiar with the landscape here. I have not had any problems. Can you please state your problem in a manner simple enough for a dunder head like me to understand? -- .''`. Stephen Powell : :' : `. `'` `- -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/ -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: ECKD driver vs DIAG driver
On 01/06/2011 01:22 PM, Stephen Powell wrote: I have recently completed work on enhancements to GNU parted to support all combinations of DASD type (CKD and FBA) DASD format (cdl, ldl, CMS non-reserved, and CMS reserved) and DASD driver (ECKD, FBA, and DIAG) supported by the Linux kernel; Excellent! -- Rich Smrcina Velocity Software, Inc. http://www.velocitysoftware.com Catch the WAVV! http://www.wavv.org WAVV 2011 - April 15-19, 2011 Colorado Springs, CO -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: ECKD driver vs DIAG driver
On Thu, 06 Jan 2011 16:10:36 -0500 (EST), Richard Troth wrote: The problem is that one cannot boot from an unpartitioned CKD disk (LDL) even though one can boot from an unpartitioned FBA disk. Partition tables are not required for other disks and bootstraps. Why should they be required for mainframe disks and bootstraps? The Linux kernel does not consider an LDL-format disk to be unpartitioned. If you format a disk with dasdfmt using -d ldl (and other appropriate parameters), then the disk has been implicitly partitioned, as far as the Linux kernel is concerned. Assuming CKD DASD, the implicit partition will begin with the fourth physical block. (The first two blocks are reserved for IPL records, the third block is the volume label.) I haven't tested your exact scenario, but here's what I have tested. I have a Linux machine that runs in a virtual machine under z/VM. It has four disks, as follows: device blockmount number special point file -- --- - 0200/dev/dasda /dev/dasda1 / 0201/dev/dasdb /dev/dasdb1 /boot 0202/dev/dasdc /dev/dasdc1 /home 0203/dev/dasdd /dev/dasdd1 swap All four of the disks are CMS reserved minidisks. All of them use the DIAG driver except 0201, which uses the ECKD driver. The boot device is 0201. Linux is started by IPL 0201 It works great. I've been doing it for years. What's the problem? (0201 has to use the ECKD driver because zipl does not support writing IPL records to a device controlled by the DIAG driver) -- .''`. Stephen Powell : :' : `. `'` `- -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: ECKD driver vs DIAG driver
Quote: Mark is correct: one automagically created partition. Worse, there is no 'fdasd' or 'fdisk' management of that partition. WORSE STILL, you *must* put the filesystem into the partition (such as it is) if you are going to boot from this disk. A filesystem in /dev/dasdx will be clobbered by the first stage of the boot loader, while a filesystem in /dev/dasdx1 is protected by the extra 8K of padding. (12K total) I checked it again this morning. The bootstrap overwrites the root inode. Well the LDL disk layout basically consists of two blocks of data (size depends on the blocksize used for formatting) Those are being used to a) label the device and b) conatin the IPL boot code (channel programs). In theory, we could mangle the IPL code into the first 512 bytes of a filesystem which is reserved on x86 as partition boot record, but that'd have a couple of downsides: a) the disk would not be labled as observed by other mainframe OSes b) this would differ from how people use Linux on other platforms where they typically do use partitions in front of the partition. Note that for LDL formatted media, you may chose to put the filesystem on the device itself (like /dev/dasdx instead of /dev/dasdx1) for volumes that you do not intend to boot from. Due to the fact that for ECKD CDL media blocks on track 0 do not have the formatted size, you cannot do the same with ECKD CDL formatted disks (filesystem corruption would be the result). with kind regards Carsten Otte IBM Linux Technology Center / Boeblingen lab -- omnis enim res, quae dando non deficit, dum habetur et non datur, nondum habetur, quomodo habenda est -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: ECKD driver vs DIAG driver
Mark is correct: one automagically created partition. Worse, there is no 'fdasd' or 'fdisk' management of that partition. WORSE STILL, you *must* put the filesystem into the partition (such as it is) if you are going to boot from this disk. A filesystem in /dev/dasdx will be clobbered by the first stage of the boot loader, while a filesystem in /dev/dasdx1 is protected by the extra 8K of padding. (12K total) I checked it again this morning. The bootstrap overwrites the root inode. This is really really sad because partitioning is really really useless otherwise. The moral equivalent in electronics is dropping resistors versus switching power supplies. As the IEEE puts it, resistance is useless. If you don't need to boot from it, use LDL and use the whole disk. Again, IFF you DON'T need to boot from it (and it's CKD), use /dev/dasdx instead of /dev/dasdx1. In any case, use DIAG wherever possible, which means use LDL wherever possible. God bless whoever in Boeblingen fixed this problem for FBA disks. You can use the pseudo-partition, or not. You can boot from them either way. -- R; Rick Troth Velocity Software http://www.velocitysoftware.com/ On Mon, Jan 3, 2011 at 14:59, Mark Post mp...@novell.com wrote: On 1/3/2011 at 10:59 AM, Samir Reddahi samir.redd...@securex.be wrote: a couple of questions: 1) Are there any disadvantages of formatting disks as ldl instead of cdl? With LDL, you only get one (automagically created) partition. With z/VM that's not such a huge problem, since disks can be made any arbitrary size. In an LPAR environment, it can cause a lot of disk to be wasted if you don't plan things out. 2) Is it true that /boot directory has to be located on a non-DIAG disk? Not to my knowledge. 3) What kind of performance improvement can I expect from a move to DIAG dasd 4) Any disadvantages of using DIAG instead of ECKD? I'll let Barton or Rob answer these two. Mark Post -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/ -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: ECKD driver vs DIAG driver
On 1/3/2011 at 10:59 AM, Samir Reddahi samir.redd...@securex.be wrote: a couple of questions: 1) Are there any disadvantages of formatting disks as ldl instead of cdl? With LDL, you only get one (automagically created) partition. With z/VM that's not such a huge problem, since disks can be made any arbitrary size. In an LPAR environment, it can cause a lot of disk to be wasted if you don't plan things out. 2) Is it true that /boot directory has to be located on a non-DIAG disk? Not to my knowledge. 3) What kind of performance improvement can I expect from a move to DIAG dasd 4) Any disadvantages of using DIAG instead of ECKD? I'll let Barton or Rob answer these two. Mark Post -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: ECKD driver vs DIAG driver
LDL is kind of a misnomer. It only means that the disk is all blocked consistently (at 4K), which is more good than bad. Last I recall, you could still boot from LDL just fine. (It is the original layout for Linux/390 CKD.) Last time I measured, I was disappointed to not find significant performance boost with the DIAG driver. That's surprising because in theory there would be a significant win with DIAG. (It is the only mode used by CMS, which gave up CCW disk I/O decades ago.) But there was no penalty. I recommend it. -- R; On Jan 3, 2011 11:01 AM, Samir Reddahi samir.redd...@securex.be wrote: We have currently a couple of DB2 machines with logical volumes on MOD-9 ECKD disks. On our DB2 production server we have mdcache enabled for the logical volumes that contains the DB2 data. The other servers have mdcache disabled. We already use the DIAG driver for the swap disks and the VDISKs. Now I'm looking if it is interesting to switch the other disks to DIAG too. Disks are currently formatted with cdl. I've read that cdl and DIAG can't work together so the disks have to be formatted to another format. We will be migrating to a new Z196 and a new DS8800, so I can use the downtime to switch the logical volume to an LV with DIAG disks a couple of questions: 1) Are there any disadvantages of formatting disks as ldl instead of cdl? 2) Is it true that /boot directory has to be located on a non-DIAG disk? 3) What kind of performance improvement can I expect from a move to DIAG dasd 4) Any disadvantages of using DIAG instead of ECKD? Best regards Samir Reddahi System Engineer | Systeem MF, AS400, DBA Operations T +32 9 235 61 09 | M +32 478 80 68 30 samir.redd...@securex.be www.securex.be - Confidentiality Notice - This communication and the information it contains is intended (a) for the person(s) or organization(s) named above and for no other person or organization, and (b) may be confidential, legally privileged and protected by law. Unauthorized use, copying or disclosure of any of it may be unlawful! If you receive this communication in error, please notify us immediately, destroy any copies and delete it from your computer system. Please consult our disclaimer on our site www.securex.eu Thank you. -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: ECKD driver vs DIAG driver
Second time I am sending this. Weird interaction with Droid GMail applet since Samir's post appeared (to GMail) to be spam (but Mark's did not, duh). Anyway, LDL is kind of a misnomer. It simply means that the disk is all blocked consistently at 4K, which is more good than bad. The only downside to LDL is that z/OS does not grok it. No great loss. (No reason z/OS could not grow up and get a driver or do SSCH to an LDL disk. Sheesh!) Last I recall, yes, you could still boot from LDL just fine. It is the original layout for Linux/390 CKD. It has been bootable all along. Last time I measured DIAG -vs- SSCH, I was disappointed to not find significant performance boost with the DIAG driver. That's surprising because in theory we would get a major win when using the DIAG driver. (It is the only mode used by CMS, which gave up CCW disk I/O decades ago.) But there was no penalty, so I recommend DIAG even if throughput comparison is a wash. -- R; Rick Troth Velocity Software http://www.velocitysoftware.com/ On Mon, Jan 3, 2011 at 10:59, Samir Reddahi samir.redd...@securex.bewrote: We have currently a couple of DB2 machines with logical volumes on MOD-9 ECKD disks. On our DB2 production server we have mdcache enabled for the logical volumes that contains the DB2 data. The other servers have mdcache disabled. We already use the DIAG driver for the swap disks and the VDISKs. Now I'm looking if it is interesting to switch the other disks to DIAG too. Disks are currently formatted with cdl. I've read that cdl and DIAG can't work together so the disks have to be formatted to another format. We will be migrating to a new Z196 and a new DS8800, so I can use the downtime to switch the logical volume to an LV with DIAG disks a couple of questions: 1) Are there any disadvantages of formatting disks as ldl instead of cdl? 2) Is it true that /boot directory has to be located on a non-DIAG disk? 3) What kind of performance improvement can I expect from a move to DIAG dasd 4) Any disadvantages of using DIAG instead of ECKD? Best regards *Samir Reddahi* System Engineer | Systeem MF, AS400, DBA Operations T +32 9 235 61 09 | M +32 478 80 68 30* **samir.redd...@securex.be* samir.redd...@securex.be [image: Securex Logo]*www.securex.be* http://www.securex.be/ [image: Footer] - Confidentiality Notice - This communication and the information it contains is intended (a) for the person(s) or organization(s) named above and for no other person or organization, and (b) may be confidential, legally privileged and protected by law. Unauthorized use, copying or disclosure of any of it may be unlawful! If you receive this communication in error, please notify us immediately, destroy any copies and delete it from your computer system. Please consult our disclaimer on our site www.securex.eu Thank you. -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: ECKD driver vs DIAG driver
LDL format was the first format used by Linux. It was quickly replaced by CDL format. The main difference is that a disk in CDL format is seen as being full by all other operating systems (VM, VSE OS/390). However, they see a LDL fromat disk as being empty. Thus if you are going to run anything other than Linux on your Z196 you must make a special effort to keep the other systems from over writing the Linux disks in LDL format. The other systems will not automatically over write a CDL format disk as they think it is already full of data that can not be deleted. DIAG works with CDL format. As DIAG was added to Linux after CDL format was in wide use, I don't know if support for LDL was added to DIAG. (It probably was.) The only advantage to LDL is you may be able to have 1 additional 4K block of Linux data on the disk. There is no reason why you can't put /boot on a DIAG disk. There are some restrictions to where you can put /boot. -- Stephen Frazier Information Technology Unit Oklahoma Department of Corrections 3400 Martin Luther King Oklahoma City, Ok, 73111-4298 Tel.: (405) 425-2549 Fax: (405) 425-2554 Pager: (405) 690-1828 email: stevef%doc.state.ok.us On 1/3/2011 9:59 AM, Samir Reddahi wrote: Disks are currently formatted with cdl. I've read that cdl and DIAG can't work together so the disks have to be formatted to another format. We will be migrating to a new Z196 and a new DS8800, so I can use the downtime to switch the logical volume to an LV with DIAG disks a couple of questions: 1) Are there any disadvantages of formatting disks as ldl instead of cdl? 2) Is it true that /boot directory has to be located on a non-DIAG disk? -- Stephen Frazier Information Technology Unit Oklahoma Department of Corrections 3400 Martin Luther King Oklahoma City, Ok, 73111-4298 Tel.: (405) 425-2549 Fax: (405) 425-2554 Pager: (405) 690-1828 email: stevef%doc.state.ok.us -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: ECKD driver vs DIAG driver
On 1/3/2011 at 08:25 PM, Stephen Frazier ste...@doc.state.ok.us wrote: DIAG works with CDL format. As DIAG was added to Linux after CDL format was in wide use, I don't know if support for LDL was added to DIAG. (It probably was.) Sorry, but that is not correct. The DIAG driver requires the disk to be in either LDL format, or CMS RESERVEd and FORMATted. Mark Post -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/