Andy/Neil, I am still having no luck with RHEL 6.3 Beta on an HP DL380 G5 with IntelX520-DA2 for the FCOE Target. I can see other targets from my server (so my basic FCOE Config is OK), but my server will not register with the name service on the Nexus switch as a target. I currently have fcoe-target chkconfig off and am manually starting it after the server is completely up. When I start the service I get no error at the console or in /var/log/messages. The only error I get is the first time I create /backingstores/block/disk1 in targetcli. At that point I get an error in /var/log/messages saying "Unable to load target_core_stgt". I will open a support case with RedHat.
Thanks, Steve ________________________________ From: Andy Grover <[email protected]> To: Nicholas A. Bellinger <[email protected]> Cc: Steve Condas <[email protected]>; Neil Horman <[email protected]>; "[email protected]" <[email protected]>; Jerome Martin <[email protected]> Sent: Thursday, May 31, 2012 2:57 AM Subject: Re: [Open-FCoE] RHEL 6.3 Beta userspace breakage for tcm_fc Hi Steve, For RHEL this bug was fixed in fcoe-target-utils-2.0rc1.fb10-4, post beta. For Fedora 17, unfortunately the analogous pkg targetcli-2.0rc1.fb11-1 doesn't have this fix either. I'll update it. It will be in the repos by Friday. In the future please do not hesitate to open bugs before fully characterizing them, or at least in parallel. Distro packages may differ from upstream (and in this case do and the differences are to blame) so we'd like to spare upstream from distro-specific issues. Thanks -- Regards -- Andy On 05/30/2012 06:55 PM, Nicholas A. Bellinger wrote: > On Wed, 2012-05-30 at 09:23 -0700, Steve Condas wrote: >> Neil, >> >> Thanks for your response. I think I have determined that the RHEL >> 6.3 Beta implementation is pretty mucked up. As best I can tell >> there is a version mismatch between the kernel drivers and the user >> space tools, so it is unsurprising that many features don't work. >> I will contact RedHat and see if I can get some patches for >> testing. >> >> > > Hello Steve + Neil, > > Just noticing this thread wrt to tcm_fc driver breakage on RHEL 6.3 > Beta code.. (CC'ing target-devel + jxm) > > I recently grabbed Fedora 17 w/ 3.3.7-1.fc17.x86_64 and was able to > load tcm_fc.ko + create /sys/kernel/config/target/fc top level > configfs group as expected, but targetcli-fb does appear to be broken > for existing /fc usage: > > Welcome to the targetcli shell:: Copyright (c) 2011 by RisingTide > Systems LLC. > > For help on commands, type 'help'. > > /> tcm_fc/ info Fabric module name: tcm_fc ConfigFS path: > /sys/kernel/config/target/fc Allowed WWNs list (free type): Fabric > module specfile: /var/lib/target/fabric/tcm_fc.spec Fabric module > features: acls Corresponding kernel module: tcm_fc /> ls tcm_fc o- > tcm_fc ..................................................... [Not > found] /> exit Global pref auto_save_on_exit=true Existing file > /etc/target/saveconfig.json backed up to saveconfig.json.backup > Configuration saved to /etc/target/saveconfig.json [root@localhost > ~]# ls -la /var/lib/target/fabric/tcm_fc.spec -rw-r--r--. 1 root root > 1130 Apr 17 13:11 /var/lib/target/fabric/tcm_fc.spec > > [root@localhost ~]# cat /sys/kernel/config/target/fc/version TCM FC > 0.3 on Linux/x86_64 on 3.3.7-1.fc17.x86_64 > > Agrover (CC'ed), any ideas on what might have caused this for RHEL > 6.3 Beta / F17 code..? > > --nab > >> Thanks again, Steve Condas >> >> >> >> ________________________________ From: Neil Horman >> <[email protected]> To: Steve Condas <[email protected]> Cc: >> "[email protected]" <[email protected]> Sent: Tuesday, May 29, >> 2012 10:47 AM Subject: Re: [Open-FCoE] (no subject) >> >> On Mon, May 28, 2012 at 03:46:31PM -0700, Steve Condas wrote: >>> All, >>> >>> I am attempting to use Open FCOE target through its >>> implementation in RHEL 6.3 Beta. I am having two issues that I >>> would like to characterize before opening a bug report with >>> Redhat: >>> >>> 1) My Nexus switch only sees my server as an initiator. I run >>> registration debug on the Nexus and I see that my server is only >>> registering with FC4 Feature 0x2 (initiator). This is different >>> than with RHEL 6.2 where it registers with 0x3 (both). I can't >>> figure out where in the FCOE code the feature is being set so >>> that I can try to debug what is going on. Can anyone who is more >>> familiar with the code point me in the right direction? >>> >>> 2) My targetcli program is throwing an error when I exit or when >>> I do an "ls" from "/". The error is that it can't find >>> auth/userid (I assume CHAP authentication from looking at the >>> code) for the ACL. Is this more likely a problem in the kernel >>> namespace or a problem with targetcli? >>> >> fc_fc4_register_provider is what adds the local server as a fcoe >> target and should trigger the apporpriate registration with the >> switch. It would seem that something may be going wrong with the >> tcm_fc module initalization preventing its registering locally as >> an fcoe provider. That may also be whats causing your strange auth >> errors (or contributing to them). Do you have any log entries >> describing problems with the insertion of the tcp_fc module? Neil >> >>> Thanks for any pointers! Steve >>> _______________________________________________ devel mailing >>> list [email protected] >>> https://lists.open-fcoe.org/mailman/listinfo/devel >>> >> _______________________________________________ devel mailing list >> [email protected] >> https://lists.open-fcoe.org/mailman/listinfo/devel > > _______________________________________________ devel mailing list [email protected] https://lists.open-fcoe.org/mailman/listinfo/devel
