I have 3 fcp tape devices defined to a Suse 9 and a Suse 10 system
running under VM. As far as I can tell, it looks like whichever system
is booted first grabs the tapes, and the system booted second writes a
message saying the tapes are already in use in lpar4.
Hi Sue,
on a z9 you can use
Our local hardware folks tell us the proper firmware is set for our two
FCP CHPID's. Since we need to be connected thru a SAN switch (assuming
that requirement is still needed) the z800 Hardware Support folks
told us to try setting a SWITCH= option on the CHPID defn in the
IOCP. We tried
Hi Arty,
I normally use evms for FCP SCSI path multipathing,
but I can't figure out how to get evms to add the requisite regions
during
the install process. I'd hate to have to pre-build the evms segments
prior
to the install process.
Have you ever considered to use multipath-tools and LVM2
1) How can I get these to come online by themselves at boot time?
Just configure your SCSI tapes with yast or manually and then run
mkinitrd to copy the current zfcp configuration to your ramdisk.
Don't forget the zipl.
2) Once the paths are online, how do the tape devices get created?
The
We have the Linux Lpar under ZVm and connected thru a McData Switch. We
see
the following on device eg. 406 as FBA but later is seen as ECKD device.
What informational source can I use to do a step by step allocation of
FBA
devices or understnd more on this topic. Thank You.
Hi Mike,
is
I need to delete it but YAST does not want to do this. In YAST I use
Harware - Fcp - select lun and delete.
LUN does not budge. Tried mkinird/zipl and rebooted - still there.
The SCSI cloning redbook uses a echo lun# unit_add to add luns. I
tried
a echo lun# unit_remove but this did not
Thank You - this worked.
Does this mean there is a problem with Yast that I should report?
Well, I usually don't use yast to remove SCSI disks from my system
so I havn't used that panel yet. But if yast promises you to delete
an FCP LUN and in fact it does not without any error message,
it's a
(I *really* wish IBM would bring back the announcement and sales manual
archives for the no-longer-marketed stuff in some form. It's really
annoying
not to be able to look this up online any more. I'd even pay for a copy
of
that older material -- it's not like it's competitive information any
Has anyone tried adding FCP tapes to linux. I was able to add FCP dasd
to Suse v8 and I'm in the process of adding them to Suse v9, but I'm
getting nothing but error messages when I try to add an FCP tape device
to either system. I suspect, but I'm not sure, that I've probably got a
problem
Is the partition limit for CDL layout on a volume still 3 in SLES 9?
Yes, ECKD CDL disk format supports only 3 partitions.
regards,
Volker
Linux on zSeries Development
IBM Deutschland Entwicklung GmbH
--
For LINUX-390
I installed a Suse Linux 9 in a z800 with fiber channel only.
The installation finished without any problem, when I IPLed Linux in LPAR
from SCSI, I got the following message :
An error was detected in partition LINU, central processor (CP) 00 is
looping due to switching between program
Sure you can:
# dasdview -j /dev/dasdb
LINUX1
# fdasd /dev/dasdb
...
Command action
...
v change volume serial
...
Command (m for help): v
Please specify the new volume serial (6 characters).
(Press ENTER if you do not want to change it)
current: LINUX1
new: test99
volume
Well, Linux native and Linux Swap partition types are independent from the
DASD layout. Both, cdl and ldl DASDs could have Linux native partitions.
The DASD layout tells you how many partitions are possible (ldl - one,
cdl - up to three). The partition type (e.g. 0x82 for Linux native) tells
you,
Is it possible to write a new volser to a pack with the system up and
running ?
Yes, it is. Just use fdasd.
There is a menu entry v change volume serial, which does the job.
Sure, fdasd will complain that your disk is in use and you could
heavily damage your system, in case you do not know,
The problem is the ramdisk on your SCSI disk. This ramdisk contains
the information, where to find your root file system (adapter devno,
WWPN and LUN). Hence you have to change this information on the
ramdisk. A simple mk_initrd will do the job on the mounted SCSI disk.
But make sure, not to
I do not see any other way. The information how to find the
SCSI root device is hard coded to the ramdisk.
In case this path is not available (e.g. the cloned guest has
a different adapter devno), the kernel is not able yet to find
its root device.
regards,
Volker
Linux on zSeries Development
SuSE SLES-9 natively supports installing onto FCP disk.
...as Robert already mentioned, in case you are using SLES9 you can
directly install to an FCP disk. There is a new panel within the
yast installation process, which asks you, if you want to install
to a SCSI disk or a DASD. In case of the
fdasd -a creates a z/OS compatible disk layout on your DASD and
writes a VTOC containing your partition information.
And in addition the first few blocks on your DASD have a
variable block size now, even if the DASD driver hides this from
you. For that reason you can not use the
Sure it has performance implications using a different file system.
Each file system has its strengths and it really depends on your
workload, which one to use. Ext3 is rock-solid because it is based
on ext3. ReiserFS saves your disk storage because of the disk space
optimization algorithm and the
Currently the SCP data field is 4096 bytes in size under z/VM and
265 bytes in size under LPAR. The format is UTF8, but currently only
the first 127 characters are supported, which means ASCII.
SCP stands for system control program. And with the SCP data field it
is possible to give some
Let me tell something about the background. Long time ago we
decided to set DS1LSTAR to zero, because it is to small for
larger partitions. The TTR field allowes us partition sizes
of max 3GB for a usual Linux ECKD geometry (12 blocks per track
and 4k block size). We did not want this restriction.
dasdview does the job
bash-2.05# dasdview -t info -f /dev/dasda
--- VTOC info
-
The VTOC contains:
1 format 1 label(s)
1 format 4 label(s)
1 format 5 label(s)
0 format 7 label(s)
Other S/390 and zSeries operating systems
Hi Tim,
you have to create a partition with fdasd after formatting:
fdasd -a /dev/dasds
regards,
Volker
I tried to format a DASD (CDL) and run into the following problem. I am
using SLES7.2 (kernel 2.4.17), s390-tools-1.1.4 and e2fsprogs-1.23-2.
First off all I used dasdfmt:
dasdfmt -f
Michael, could you please send me the following information offline?
I would like to have a look into the problem.
dasdview -b 0 -s 3b -2 -f /dev/dasde (first 3 blocks of track 0)
dasdview -b 1t -s 6b -2 -f /dev/dasde (first 6 blocks of track 1)
dasdview -b 2t -s 1b -2 -f /dev/dasde (first
Hi Dougie
Excuse me being pedantic here, but those instructions seem to be for cdl
formatted (2.4 kernel) not ldl formatted (2.2 kernel) dasd.
The subject of this thread was for 2.2.x kernels, so isn't the only option
in this case the OFFLINDR stuff or OEM utilities, again.
...more information and examples about using z/OS DFSMSdss/hsm
with Linux could be found here
http://www.storage.ibm.com/software/sms/dss/linux_dump_restore.html
and
http://www.storage.ibm.com/software/sms/hsm/linux_dump_restore.html
regards, Volker
Volker,
For the future, how should broken links be reported? To you directly?
Via
the mailing list as Rick did? Some other way?
Mark Post
Mark,
please report broken links or other things related to our DeveloperWorks
website to [EMAIL PROTECTED]. This should be the fastest way.
This
Yep, they're broken. No address to report them, either. Sigh.
Mark Post
A couple of the glibc patches (the TARred-up diffs)
notably, glibc-2.2.4-s390-3 and 2.
Yes, I have found three broken glibc links and they are fixed now.
Rick, thank you for reporting this.
regards, Volker
ldl cdl:
Cyl 0 Head 0 Track 2: IPL record
ldl cdl:
Cyl 0 Head 0 Track 3: Label
cdl:
Cyl 0 Head 0 Tracks 4-5: DSCBs to make it look like one through four data
sets
this is not quite correct :)
The Linux partition table in cdl case is hidden in a 390-like
volume table (VTOC) in the second
Mark,
not necessarily. This change has no real impact to Linux. The only visible
difference is the data set name. Here we have now PART0001 up to PART0003
instead of PART up to PART0002.
All other VTOC information (e.g. partition start and size) will not change.
If you use these data set
Port
DFSMSdss requires a volume serial to dump a dasd. The volume initialized
as
Linux volume does not have a volser. How does DFSMSdss recognize the
volume
for dumping?
Thanks.
IC Liang.
-Original Message-
From: Volker Sameske [mailto:[EMAIL PROTECTED]]
Sent: Wednesday
[EMAIL PROTECTED]
ARIST.EDU
21.01.02 13:18
Please respond
to Linux on 390
Port
Volker Sameske wrote:
If you already use DFSMSdss to backup your OS/390 disks, you can use
32 matches
Mail list logo
