Okay, I asked the guy who installed it tools and kernel modules are from oel 5.5. We changed the configuration to a dell supported configuration *new network cards *jumbo fromes *multipathing enabled
now we do not have the problem anymore, despite I can not explain why. Thx for the help anyway On 26 Aug., 22:46, Mike Christie <micha...@cs.wisc.edu> wrote: > On 08/26/2010 03:04 AM, bina...@web.de wrote: > > > Hello, > > > we are using open iscsi downloaded from here :http://www.open- > > iscsi.org/bits/open-iscsi-2.0-871.tar.gz > > Are you using the userspace and kernel modules that come with that > tarball, or just the tools? > > Why don't you use the tools and/or kernel modules that come with oel? > > Do you see the problem with only the tarball kernel modules or do you > also see it with the kernel modules that come with oel? > > Has Oracle tried to debug this already or has equallogic? > > > with oracle Enterprise Linux 5.5: > > > [r...@rac1 ~]# uname -a > > Linux rac1 2.6.18-194.el5 #1 SMP Mon Mar 29 22:10:29 EDT 2010 x86_64 > > x86_64 x86_64 GNU/Linux > > [r...@rac1 ~]# iscsiadm -V > > iscsiadm version 2.0-871 > > > This is as two node RAC with ASM and about 7 Luns each Database one. > > During Performance Tests > > we get a stale or broken session after a couple of hours running the > > Is there anything in the target logs? > > > test. We can then copy > > data fast to the SAN - some gigybyte in some seconds but can read it > > only slow - 20 MB > > What is the normal throughput? > > When there is no problem can you do > cat /sys/class/scsi_host/hostX/host_busy > And can you run that again when there is a problem and check if there is > a difference? > > > > > in more the one minute. the data will be read correct but it is slow. > > After logoff and > > login for just the one session the session is fast as before. We can > > reproduce the > > probelm with the next performance test. Unfortunally we have no error > > messages in the > > dmesg or /var/log/messages files. I checked this it is just slow no > > messages. Since we want > > to keep it simple we have no multipathing or bonding only one > > interface with one cable. > > > Since the perfromance tests do not finish we have some time pressure > > this is a new > > Dell equalogic ps6000 Hardware and server. Can anybody tell us how to > > fix the problem > > or what to try. Is there an explaination why only read performance is > > affected. With > > a RAC test it is possible that one node can still read fast ant the > > other not unter the > > sesison gets a now login. Are there any other relevant logfiles or > > debug modes etc? > > You can turn on debugging by writing 1 to the iscsi modules debugging > sysfs files. > > /sys/module/iscsi_tcp/paramters/debug_iscsi_tcp > /sys/module/libiscsi_tcp/paramters/debug_libiscsi_tcp > /sys/module/libiscsi/paramters/debug_libiscsi_* > > This will make lots of output get written to /var/log/messages -- You received this message because you are subscribed to the Google Groups "open-iscsi" group. To post to this group, send email to open-is...@googlegroups.com. To unsubscribe from this group, send email to open-iscsi+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/open-iscsi?hl=en.
