>>> On 9/8/2011 at 03:43 AM, Lu GL Gao <lu...@cn.ibm.com> wrote: -snip- > The following is what I did: > (1)use YAST2 to add 2 zfcp disks(actually they are one SCSI disk).
I don't know what this means, exactly. > successful But I'm glad it was successful. > (2)use mkinitrd command. The response messages are different with those of > SUSE 10, I'm not sure if they are correct. > LXCPOA:/ # mkinitrd > > Kernel image: /boot/image-2.6.32.12-0.7-default > Initrd image: /boot/initrd-2.6.32.12-0.7-default > Root device: /dev/disk/by-path/ccw-0.0.0200-part1 (/dev/dasda1) (mounted > on / as ext3) > Kernel Modules: jbd mbcache ext3 dasd_mod dasd_eckd_mod > Features: block dasd resume.userspace resume.kernel > 25502 blocks Unless you're changing something about your root file system, regenerating the initrd is usually not needed. In any case, the output looks fine. > (3)use zipl command. The response messages are also different with those of > SUSE 10, I'm not sure if they are correct. > LXCPOA:/ # zipl > Using config file '/etc/zipl.conf' > Building bootmap in '/boot/zipl' > Building menu 'menu' > Adding #1: IPL section 'SLES11_SP1' (default) > Adding #2: IPL section 'FailsafeV1' > Adding #3: IPL section 'ipl' > Preparing boot device: dasda (0200). > Done. If you don't update your initrd, you don't need to re-run zipl (in this particular context). But again, the output looks fine. > (4)use lsscsi command to show available zfcp disks. And check their > information. successful(I think) > LXCPOA:/ # lsscsi > [0:0:0:1073758208]disk IBM 2107900 3.44 /dev/sda > [1:0:0:1073758208]disk IBM 2107900 3.44 /dev/sdb Looks normal to me. -snip- > (5)At this point, there is no multipath.conf file in system, so I created > new one. > LXCPOA:/etc # cat multipath.conf > multipaths{ > multipath{ > wwid 36005076308ffc6210000000000000000 > alias mpvol1 > } > } > > (6)I want to enable new alias, so I enable multipath again. But nothing is > shown, is there a error? > LXCPOA:/etc # multipath > LXCPOA:/etc # This command does _not_ "enable multipath." chkconfig multipathd on, followed by /etc/rc.d/multipathd start will do that. Once that is done, run "multipath -ll" to see if things look better. > (7)use YAST2 to have a file system created on it and mounted. successfuly > > (8)check new directory. I don't know why my alias to this disk is not > used!!! > LXCPOA:/ # df -h > Filesystem Size Used Avail Use% Mounted on > /dev/dasda1 2.3G 438M 1.8G 20% / > devtmpfs 497M 204K 497M 1% /dev > tmpfs 497M 100K 497M 1% /dev/shm > /dev/dasdb1 2.3G 1.9G 293M 87% /usr > /dev/dasdc1 2.3G 1.4G 813M 63% /usr/share > /dev/dasdd1 2.3G 195M 2.0G 9% /var > /dev/mapper/36005076308ffc6210000000000000000_part1 > 778G 197M 738G 1% /vol1 > 1244.70 2.54 9955.06 6778 26563184 Not every tool reports the same information. Confusing and concerning, I know. Personally, I would welcome support requests from customers who feel the same. > (9)I upload a 1G file by FTP to new directory to check if IO load is > balanced among 2 paths. > Is balance sucessful? Why the "tps" value for dm-0 and dm-1 is so > different? > LXCPOA:/ # iostat > Linux 2.6.32.12-0.7-default (LXCPOA) 09/07/2011 _s390x_ > > avg-cpu: %user %nice %system %iowait %steal %idle > 2.62 0.01 2.66 1.02 0.36 93.33 > > Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn > dasda 2.66 116.17 98.00 317808 268112 > dasdb 1.22 121.49 0.27 332352 744 > dasdc 1.15 83.30 0.97 227896 2664 > dasde 0.06 1.74 0.00 4764 0 > dasdd 0.97 13.11 15.88 35856 43432 > dasdf 0.02 0.46 0.00 1248 0 > dasdg 0.02 0.46 0.00 1248 0 > dasdh 0.02 0.46 0.00 1248 0 > sda 8.68 17.53 5090.43 47962 13926104 > dm-0 17.07 17.12 10194.13 46834 27888488 > sdb 8.42 0.52 5103.70 1432 13962384 > dm-1 1274.58 2.49 10194.13 6818 27888480 The first set of output from iostat is from the time of system startup until now. You might be better off running iostat continuously (iostat 2 or iostat 5) and watching how things are working. In your multipath.conf file you didn't specify a path_grouping_policy. I don't know what the default is. It may or may not be one that provides load balancing. It may be providing failover. 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/