[stgt] scsi_id not valid in iscsi node

lingli tang tanglingli001 at gmail.com
Mon Jun 23 05:09:14 CEST 2014


so after commit 8141d67b. the tgtd is not Backward compatible?
and we have to use tgtd before commit 8141d67b for compatible with
multipath/udev of rhel6.2
or rebuild a different multipath/udev in  rhel6.2?

2014-06-15 14:23 GMT+08:00 Or Gerlitz <ogerlitz at mellanox.com>:
> On 03/06/2014 09:09, Apollon Oikonomopoulos wrote:
>>
>> On 14:43 Mon 02 Jun     , Nicholas A. Bellinger wrote:
>>>
>>> Forwarding to the correct stgt-devel list..
>>>
>>> On Fri, 2014-05-30 at 17:38 +0800, lingli tang wrote:
>>>>
>>>> [...]
>>>> 1. [...] cpu(%user,%sys,%iowait,%idle),memory(Total,used, free,
>>>>
>>>> buffer, cache), IOPS(read request, write request), Througthput(read
>>>> kb, write kb),
>>>>
>>>> we use tgtd 1.0.47 to export a disk and use iscsiadm to load it.
>>>> but the ID_SERIAL of this disk is not the same with scsi_id of export
>>>> disk.
>>>> if we use tgtd 1.0.24 it is the same.
>>>> after iscsiadm log in, we check the ID_SERIAL of this disk using
>>>> “udevadm info --query=all --path=/block/sdb”:
>>>>      E: ID_SERIAL_RAW=360000000000000000e001000c00000e0
>>>>      E: ID_SERIAL=360000000000000000e001000c00000e0
>>>>      E: ID_SERIAL_SHORT=60000000000000000e001000c00000e0
>>>>
>>>> but with tgtd 1.0.24, it is correct:
>>>>      E: ID_SERIAL_RAW=1test1-iscsi-iser
>>>>      E: ID_SERIAL=1test1-iscsi-iser
>>>>      E: ID_SERIAL_SHORT=test1-iscsi-iser
>>>>
>>>> test environment rhel6.2
>>
>> We were debugging the exact same thing yesterday, after our persistent
>> links under /dev stopped working, so here's the short version of the
>> story:
>>
>> Commit 6402eb8 added an NAA local designator (WWN) field to tgt's 0x83
>> VPD (the Device Identification page), and commit 8141d67b further added
>> an extended WWN. All three identifiers are sent out in tgt's 0x83 VPD,
>> but udev's scsi_id helper always picks the binary WWN by preference and
>> uses it as ID_SERIAL, i.e. ignores the ASCII T10 (ID_SERIAL) identifier
>> present in the VPD. If you use sg_inq from the sg3-utils package you
>> will get both.
>>
>> I'm not sure tgt can do much more than make both NAA identifiers
>> optional, the issue seems to lie with udev at this point.
>>
>>
>
> I added Dennis who is the author of commit 8141d67b -- he made the following
> comment: "[...] This is most likely an issue with udev. Most of the udev
> tools look at Inquiry VPD 0x83 before looking at VPD 0x80.  With the NAA
> pages now defined in VPD 0x83, the tools are using that information instead
> of VPD 0x80."
--
To unsubscribe from this list: send the line "unsubscribe stgt" in
the body of a message to majordomo at vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html



More information about the stgt mailing list