Re: Root almost filled on 3390-3
Gordon and all the others who replied, thanks for all the information you provided. I think I've enough information now to keep on with my Linux installation. Werner "Wolfe, Gordon W" wrote: > > Werner, > > I'm glad I was able to help you find your storage problem. I remember when sonmeone >showed me the "du" command about a year and a half ago how useful it was to me. > > As far as LVM is concerned, there are others on this list who are more qualified to >speak about LVM than I am. I'm sure this question has been answered in this list >before. You might check the list archives at >http://www.marist.edu/htbin/wlvindex?linux-vm and do a search (at the bottom of the >page) on LVM. Be patient. The search can take a while. > > Having said that, I do seem to recall that someone posted the fact that even the >boot disk can be on a logical volume. Probably any mount point can be on a logical >volume. > > Here at Boeing, we use LVM volumes mostly for user data, /home for example, or for >Oracle databases, so that we can use more than one physical volume for a mount point. > We also keep the Oracle code in logical volumes so we can have more than one mount >point on a minidisk. For our general purpose (non-Oracle, non-WebSphere) linux >systems, we create the following minidisks: > > 292 mounted as / (boot disk) > 293 V-disk for swap > 294 mounted shared read-only as /usr > 295 mounted as /home > > For boot purposes, Linux needs to have (at least) access to /boot, /bin, /sbin and >/etc on the boot disk until it is up far enough to mount other filesystems. It may >also need /var or /tmp. It definitely does not need /usr, /opt or /home unless >you've done something really radical to change your system. > > One more thing. I wouldn't get too enamored of LVM just yet. I was in a meeting >yesterday and one of the Unix gurus we work with on Linux/390 (and who is usually >pretty knowlegeable about these things) mentioned that LVM is going to be sunsetted. >It is rumored that Sistina will not be enhancing it beyond the 2.4 kernel, only >providing basic maintenance. This same person said that Linus won't be putting LVM >into 2.6 when it comes out. LVM will apparently be replaced by something similar but >more capable from IBM, and that this new filesystem is already in the 2.4.17 kernel. >There was an IBM rep there at the meeting and he seemed to know about this change as >well. We've put all our expansion of LVM on hold until we find out if this rumor is >true and (if so) what the replacement is and how you work with it. > > Perhaps others on the list can expand on this rumor and tell me if I'm just blowing >smoke and spreading FUD. > > They say there are three signs of stress in your life. You eat too much junk food, >you drive too fast and you veg out in front of the TV. Who are they kidding? That >sounds like a perfect day to me! > Gordon Wolfe, Ph.D. (425)865-5940 > VM & Linux Servers and Storage, The Boeing Company > > > -- > > From: Werner Kuehnel > > Reply To: Linux on 390 Port > > Sent: Wednesday, February 5, 2003 3:42 AM > > To: [EMAIL PROTECTED] > > Subject: Re: Root almost filled on 3390-3 > > > > Wolfe, > > thanks for your reply. The du command was a great help in finding the culprit. > > There were two files in /tmp from the installation of a DB2 fixpack which used > > up a lot of space. Now we are at 40% and 870MB. > > However, I'd still like to know which directories can be moved to LVM managed > > space (only those which will not be accessed until LVM initialization). Is this > > usual to have parts of root fs in LVM space? In my opinion /var, /usr and /tmp > > would be good candidates, wouldn't they? > > Sorry for the dumb questions, but I'm just starting with Linux. > > > > Werner > > > > "Wolfe, Gordon W" wrote: > > > > > > What's on your root? And what's on the other disks? Do a "df" and tell us the >results. > > > > > > I've been running SLES7 for a year with root on a 750-cylinder minidisk and /usr >on a dedicated 3390-3 shared read-only, /home on a 500-cylinder minidisk and swap on >a v-disk. I've never filled up root. (Okay, so there was that incident with the TSM >client logs...)> > > > > > > Check your logs to be sure they aren't filling up space in /var/log. > > > > > > cd /var/log > > > du --max-depth=1 -h > > > > > > find out where your big usages a
Re: Root almost filled on 3390-3
Sorry, it should have been Gordon, Werner Kuehnel wrote: > > Wolfe, -- Werner Kuehnel IMD GmbH (Mannheimer Versicherung) Mannheim - Germany
Re: Root almost filled on 3390-3
Wolfe, thanks for your reply. The du command was a great help in finding the culprit. There were two files in /tmp from the installation of a DB2 fixpack which used up a lot of space. Now we are at 40% and 870MB. However, I'd still like to know which directories can be moved to LVM managed space (only those which will not be accessed until LVM initialization). Is this usual to have parts of root fs in LVM space? In my opinion /var, /usr and /tmp would be good candidates, wouldn't they? Sorry for the dumb questions, but I'm just starting with Linux. Werner "Wolfe, Gordon W" wrote: > > What's on your root? And what's on the other disks? Do a "df" and tell us the >results. > > I've been running SLES7 for a year with root on a 750-cylinder minidisk and /usr on >a dedicated 3390-3 shared read-only, /home on a 500-cylinder minidisk and swap on a >v-disk. I've never filled up root. (Okay, so there was that incident with the TSM >client logs...) > > Check your logs to be sure they aren't filling up space in /var/log. > > cd /var/log > du --max-depth=1 -h > > find out where your big usages are by CD'ing into various directories and trying the >du command above. I found that installing IBMJava2-1.4 used up about 350 megs! > > Do you have Oracle installed? Websphere? Those can be HUGE! > > They say there are three signs of stress in your life. You eat too much junk food, >you drive too fast and you veg out in front of the TV. Who are they kidding? That >sounds like a perfect day to me! > Gordon Wolfe, Ph.D. (425)865-5940 > VM & Linux Servers and Storage, The Boeing Company > > > -- > > From: Werner Kuehnel > > Reply To: Linux on 390 Port > > Sent: Tuesday, February 4, 2003 3:49 AM > > To: [EMAIL PROTECTED] > > Subject: Root almost filled on 3390-3 > > > > I've installed SLES7 (Beta version) onto one (of three) 3390-3 dasd. The root > > filesystem is now filled up to 92%. I'd like to have more freespace and wonder > > how to do this. > > Books say that root filesystem under LVM is not recommended. Are there at least > > some directories I can move (of course on the fly) from root filesystem to LVM > > space? > > Are there any recommendations/experiences how to split up root fs to more than > > one 3390-3 volume? > > Any hints are very welcome. > > > > Werner > > -- > > > > Werner Kuehnel > > IMD GmbH (Mannheimer Versicherung) > > Mannheim - Germany > > > > -- Werner Kuehnel IMD GmbH (Mannheimer Versicherung) Mannheim - Germany
Root almost filled on 3390-3
I've installed SLES7 (Beta version) onto one (of three) 3390-3 dasd. The root filesystem is now filled up to 92%. I'd like to have more freespace and wonder how to do this. Books say that root filesystem under LVM is not recommended. Are there at least some directories I can move (of course on the fly) from root filesystem to LVM space? Are there any recommendations/experiences how to split up root fs to more than one 3390-3 volume? Any hints are very welcome. Werner -- Werner Kuehnel IMD GmbH (Mannheimer Versicherung) Mannheim - Germany
Re: SILO refuses to work
Thanks, Mark, for your reply, very interesting and informative. Werner "Post, Mark K" wrote: > > Werner, > > Look in /usr/src/linux/Documentation/devices.txt. '58 block' is "Reserved > for logical volume manager." You've basically run into a restriction > imposed by silo, zilo, and zipl. All the files that are used to define an > "IPL set" (for lack of a better term) _must_ reside on the same physical > volume that is going to be the IPL volume. The way it checks for this is by > comparing the major and minor device nodes for that volume against that of > all the files used. In your case, 58/0 does not match 94/0. Bang, you're > dead. > > You're going to need to create a non-LVM file system on the volume from > which you wish to IPL. Or, if you've got VM, you can IPL from the virtual > reader, freeing you from any restrictions imposed by silo, zilo, and zipl. > > Mark Post > > -Original Message- > From: Werner Kuehnel [mailto:[EMAIL PROTECTED]] > Sent: Tuesday, November 26, 2002 3:28 AM > To: [EMAIL PROTECTED] > Subject: Re: SILO refuses to work > > > Mark, > > I think your guess is correct. I'm curious as to why you want to make > your > > root file system an LVM volume, though. Do you really think you'll need > to > > expand it that much, instead of just adding additional file systems on > > various mount points? > > > to be honest, I just wanted to try out the LVM. And since our AIX guys has > also > a volume group for the root filesystem I thought that would be the normal > way to > define all available space in volume groups and logical volumes. I'm just > starting with LINUX and try out things to get familiar with it. > Can I have a mixture of non-LVM space and LVM-space? Is it not usual to have > the > root fs in an VG/LV? > > How can I find out what device 58/0 is (/boot/ipleckd.boot is not on device > (94/0) but on (58/0)) ? > > > In any case, one way to test your assumption is to get rid of the > > root-on-LVM setup and try it with just a "normal" ext2 file system on > > /dev/dasda. I strongly suspect that will work for you. > > VGROOT spans over 2 disks, but it includes several LVs: > /dev/vgroot/lvroot vgroot 2048 1 > /dev/vgroot/lvusrvgroot 2048 1 > /dev/vgroot/lvvarvgroot 256 1 > /dev/vgroot/lvhome vgroot 128 1 > > 1 disk has a capacity of approx. 2,5 GB, so anything (including /boot, > excluding > /usr, /var, /home) should be on the first disk, shouldn't it? > > -- > > Werner Kuehnel > IMD GmbH (Mannheimer Versicherung) > Mannheim - Germany -- Werner Kuehnel IMD GmbH (Mannheimer Versicherung) Mannheim - Germany
Re: SILO refuses to work
Mark, > I think your guess is correct. I'm curious as to why you want to make your > root file system an LVM volume, though. Do you really think you'll need to > expand it that much, instead of just adding additional file systems on > various mount points? > to be honest, I just wanted to try out the LVM. And since our AIX guys has also a volume group for the root filesystem I thought that would be the normal way to define all available space in volume groups and logical volumes. I'm just starting with LINUX and try out things to get familiar with it. Can I have a mixture of non-LVM space and LVM-space? Is it not usual to have the root fs in an VG/LV? How can I find out what device 58/0 is (/boot/ipleckd.boot is not on device (94/0) but on (58/0)) ? > In any case, one way to test your assumption is to get rid of the > root-on-LVM setup and try it with just a "normal" ext2 file system on > /dev/dasda. I strongly suspect that will work for you. VGROOT spans over 2 disks, but it includes several LVs: /dev/vgroot/lvroot vgroot 2048 1 /dev/vgroot/lvusrvgroot 2048 1 /dev/vgroot/lvvarvgroot 256 1 /dev/vgroot/lvhome vgroot 128 1 1 disk has a capacity of approx. 2,5 GB, so anything (including /boot, excluding /usr, /var, /home) should be on the first disk, shouldn't it? -- Werner Kuehnel IMD GmbH (Mannheimer Versicherung) Mannheim - Germany
SILO refuses to work
Although reading and searching huge amounts of threads I can't figure out how to come around this problem. After installation of SLES for S/390 YaST tries to run SILO. It doesn't work, because YaST does not fill the -d option with the DASD (at least this is my opinion): The command chroot /mnt /sbin/silo -d -t2 resulted in an error. The command gave the following output: silo.c (line:280) 'stat (name, &dst)' returned 2='No such file or directory' o->ipldevice set to -t2 Anyway, I tried to run SILO manually and got: SuSE Instsys suse:/root # chroot /mnt /sbin/silo -d /dev/dasda -F /etc/silo.conf -t2 o->ipldevice set to /dev/dasda o->conffile set to /etc/silo.conf Testonly flag is now 0 Testlevel is set to -2 IPL device is: '/dev/dasda' /boot/ipleckd.boot is not on device (94/0) but on (58/0) silo.c (line:366) 'verify_file (o->bootsect, dev)' returned 22='Invalid argument' bootsector is: '/boot/ipleckd.boot'Usage: This is my silo.conf file: ipldevice=/dev/dasda image=/boot/image bootsect=/boot/ipleckd.boot map=/boot/boot.map root=/dev/dasda1 parmfile=/boot/parmfile testlevel=-2 readonly This is /proc/dasd/devices: 3397(ECKD) at (94:0) is dasda:active at blocksize: 4096, 601020 blocks, 2347MB 3398(ECKD) at (94:4) is dasdb:active at blocksize: 4096, 601020 blocks, 2347MB 3399(ECKD) at (94:8) is dasdc:active at blocksize: 4096, 601020 blocks, 2347MB 1097(ECKD) at (94:12)is dasdd:active at blocksize: 4096, 601020 blocks, 2347MB How can I find out what device (58/0) is where it finds /boot/ipleckd.boot? Volume group VGROOT is spanned over 2 devices, 3397 and 3399. Could that be the problem, that /mnt/boot is on the other device than 3397? Any help is very welcome. Werner -- Werner Kuehnel IMD GmbH (Mannheimer Versicherung) Mannheim - Germany
Re: Function Keys
Thanks for all your replies, I was able to proceed with the installation, however had another problem. I'll open a new thread. Werner "Hall, Ken (ECSS)" wrote: > > Doesn't even seem to need the CTRL all the time. I just did an install and for some >reason the F-keys didn't work with yast using xterm, so I tried "F 1", "F 2" (just >press F, then the number), and > it worked. F11 was F "dash" (to the right of the 0), and F12 was "F =" > > > -Original Message- > > From: Werner Kuehnel [mailto:[EMAIL PROTECTED]] > > Sent: Wednesday, November 20, 2002 3:23 AM > > To: [EMAIL PROTECTED] > > Subject: Re: [LINUX-390] Function Keys > > > > > > Thanks for your replies. cntrl-F and 0 works. But what shall > > I hit for PF11 or > > PF12? I don't know if this is ever required during the > > installation, or do they > > just use PF1-PF10? > > > > Werner > > > > David Boyes wrote: > > > > > > Hit cntrl-F, release the keys, and then hit 0. This works > > consistently > > > for all pf keys (cntrl-F and 1 thru = for PF1-PF12) on all terminal > > > types (including my venerable ADM1). > > > > > > -- db > > > > > > David Boyes > > > Sine Nomine Associates > > > > Werner Kuehnel > > IMD GmbH (Mannheimer Versicherung) > > Mannheim - Germany > > -- Werner Kuehnel IMD GmbH (Mannheimer Versicherung) Mannheim - Germany
Re: Function Keys
Thanks for your replies. cntrl-F and 0 works. But what shall I hit for PF11 or PF12? I don't know if this is ever required during the installation, or do they just use PF1-PF10? Werner David Boyes wrote: > > Hit cntrl-F, release the keys, and then hit 0. This works consistently > for all pf keys (cntrl-F and 1 thru = for PF1-PF12) on all terminal > types (including my venerable ADM1). > > -- db > > David Boyes > Sine Nomine Associates Werner Kuehnel IMD GmbH (Mannheimer Versicherung) Mannheim - Germany
Function Keys
I'm currently installing again SLES for S/390. For opening a session to LINUX I use Hummingbird Exceed Hostexplorer 6.2.0.0 (i.e. telnet with VT220). Unfortunately my function keys do not work, just PF1-PF4 are acting as expected. There is a keyboard mapping display to define specific actions/keys to the keys, but I couldn't find nothing that relates to PF5-PF12. So I'm currently stuck at the point where I can start the FTP from YaST, but it requires a PF10 key. Any help is very appreciated, Werner Kuehnel IMD GmbH (Mannheimer Versicherung) Mannheim - Germany
Re: Device busy
No, just one CEC. I also believe that a POR would do the job, but I fear I won't get the time for a POR :-( I'd like to know if this is a consequence of dynamically changing the access to the devices and OS/390 2.10 or JES3 doesn't really give them free, although they're offline. Werner Kuehnel IMD GmbH (Mannheimer Versicherung) Mannheim - Germany "McKown, John" wrote: > > Ouch! That hurts. I'm at a loss. I am fairly sure that a POR of the box will > clear the condition. But that is very extreme and likely to get other people > a bit upset at you . Do you have more than one CEC? Could another CEC > have the device "tied up"? > > -- > John McKown > Senior Technical Specialist > UICI Insurance Center > Applications & Solutions Team > +1.817.255.3225 > --
Re: Device busy
Hi John, just tried it, not even with a RESET CLEAR it works. Werner "McKown, John" wrote: > > Werner, > I got something like that on an OS/390 IPL attempt. The same message, and > the devices in question were "off line" to all other LPARs. I then did a > "reset clear" on the LPAR that I was attempting to IPL, then IPL'ed it > again. This time it worked! It appears that the microcode assumed that the > IPL was still being attempted, even after the error message was displayed. > The "RESET CLEAR" cleared up the problem. > > -- > John McKown > Senior Technical Specialist > UICI Insurance Center > Applications & Solutions Team > +1.817.255.3225 > > > -Original Message- > From: Werner Kuehnel [mailto:werner.kuehnel@;mannheimer.de] > Sent: Monday, November 11, 2002 5:05 AM > To: [EMAIL PROTECTED] > Subject: Device busy > > > I installed Linux (SLES for S/390) onto 3 3390-3 volumes some months ago and > got it started from disk. When trying to restart last week I discovered that > the 3 volumes were no longer accessible by the Linux LPAR, sometime the > configuration changed. So I changed the IODF/IOCDS dynamically to allow the > Linux LPAR access to only the 3 volumes and the OSA and OSA-Express devices > (chpids are of course shared by EMIF). After starting the LOAD at HMC the > following message appear: "The load control unit and device is busy. The > load control unit and device may be shared by another system." The 3 volumes > are offline and has no logical paths in all running LPARs. All other LPARs > are deactivated. I've searched all archives, fora and redbooks, well, there > are some hits, mostly with IPL from tape, but couldn't find yet any hint how > to overcome this problem. > > I'd really appreciate any help. > > Thanks, > Werner > -- > > _ > > Werner Kuehnel > IMD GmbH (Mannheimer Versicherung) > Mannheim - Germany > > "Besuchen Sie uns im Internet: http://www.mannheimer.de"; -- _ Werner Kuehnel IMD GmbH (Mannheimer Versicherung) Mannheim - Germany "Besuchen Sie uns im Internet: http://www.mannheimer.de";
Device busy
I installed Linux (SLES for S/390) onto 3 3390-3 volumes some months ago and got it started from disk. When trying to restart last week I discovered that the 3 volumes were no longer accessible by the Linux LPAR, sometime the configuration changed. So I changed the IODF/IOCDS dynamically to allow the Linux LPAR access to only the 3 volumes and the OSA and OSA-Express devices (chpids are of course shared by EMIF). After starting the LOAD at HMC the following message appear: "The load control unit and device is busy. The load control unit and device may be shared by another system." The 3 volumes are offline and has no logical paths in all running LPARs. All other LPARs are deactivated. I've searched all archives, fora and redbooks, well, there are some hits, mostly with IPL from tape, but couldn't find yet any hint how to overcome this problem. I'd really appreciate any help. Thanks, Werner -- _________ Werner Kuehnel IMD GmbH (Mannheimer Versicherung) Mannheim - Germany "Besuchen Sie uns im Internet: http://www.mannheimer.de";
