On Tue, Nov 4, 2008 at 7:19 PM, Peter Cudhea <[EMAIL PROTECTED]> wrote: > Thanks for pointing me back to the earlier messages in the thread. > When the thread originally came out, I was not involved in iSCSI. > > The earlier messages said that you took snoop traces and saw the HP > initiator sending a SendTargets response to the target and then getting > back a Target address of (for example) 192.168.1.70,1 . And you said > that the HP initiator chokes on this format. > > This format is legal according to the iSCSI RFC 3720 (e.g. > http://www.faqs.org/rfcs/rfc3720.html). See in particular section 12.8. >> TargetAddress=domainname[:port][,portal-group-tag] >> > The source code for computing the SendTargets response in the > OpenSolaris iSCSI target is available online. As far as I can tell, > the relevant section that creates the SendTargets response is function > add_target_address in util.c: > http://src.opensolaris.org/source/xref/onnv/onnv-gate/usr/src/cmd/iscsi/iscsitgtd/util.c#add_target_address > As you describe, this code always uses the form <ip-address>,<tpgt> , > which is the form that you say the HP initiator is having trouble with. > So as far as OpenSolaris is concerned, there is no current workaround, > at least at the SendTargets level. > > An ideal solution would be for the HP initiator to be upgraded to > recognize this legal form of TargetAddress response. > > As an alternative potential workaround, have you done any > experimentation with either static discovery or iSNS discovery? It is > slightly possible that these alternate discovery methods might go > through different code paths in the HP initiator that do not object to > the OpenSolaris response.
According to http://docs.hp.com/en/T1452-90011/ch04s01.html (item 4), HP-UX supports static target configuration with TPG specified, so there's some hope (albeit a small one, I believe - I'd be surprised if negotiation code is target configuration type-specific). Regards, Andrey > > Peter > > Sean Alderman wrote: >>> There have been numerous OpenSolaris iSCSI target bug >>> fixes since build >>> 77. Whenever you get a chance, it would be >>> interesting to retry this >>> experiment with a more recent version of OpenSolaris. >>> >>> Peter >>> >>> >> >> Peter, >> Please see my first post in the thread... I started on OpenSolaris >> 2008.05 (svn_86) and recently ran a pkg image-update to bring the server up >> to svn_99. Still no luck. >> >> Unfortunately, our HP-UX systems are a significant reason behind our >> attempt to migrate from locally attached storage arrays to iSCSI from a/many >> X4500s. In our test environments we find the need to relocate entire >> arrays, while preserving the volume groups on them, from system to system. >> iSCSI and the X4500 would be a perfect replacement... if it worked (on the >> HP-UX side). I'm not a big HP-UX fan myself, but we do have a good bit of >> it here. My X4500 goes back to Sun this week, it's an eval unit. >> >> >>> I just loaded an IA64 system with HP-UX specifically >>> to test iSCSI against our Sun Fire X4500 running >>> OpenSolaris (svn_99). >>> >>> I see the same issue. >>> >>> I've had no issues with Linux, Solaris or Windows >>> initiators connecting to the same server. >>> > _______________________________________________ > storage-discuss mailing list > [email protected] > http://mail.opensolaris.org/mailman/listinfo/storage-discuss > _______________________________________________ storage-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/storage-discuss
