search

open-iscsi / rtslib-fb

Python library for configuring the Linux kernel-based multiprotocol SCSI target (LIO)
Apache License 2.0
73 stars 90 forks source link

targetcli problem with normalize_wwn() in utils.py #139

Open Aleksey-Maksimov opened 6 years ago

Aleksey-Maksimov commented 6 years ago

Hello

In rtslib/utils.py there is a function normalize_wwn():

https://github.com/open-iscsi/rtslib-fb/blob/b2ec3746fb772aa3ff8b8853965292ca7dc2d7b1/rtslib/utils.py#L371

There is a line checking WWN/WWPN here:

'naa': lambda wwn: re.match(r"naa\.[125][0-9a-fA-F]{15}$", wwn),

In my opinion, this line should be changed as follows:

'naa': lambda wwn: re.match("naa\.[0-9a-fA-F]{16}$", wwn),

Otherwise, the targetcli utility in Debian 9 has a problem.

This targetcli utility uses the package https://packages.debian.org/ru/stretch/python3-rtslib-fb, in which the normalize_wwn() function is called from the script /usr/lib/python3/dist-packages/rtslib_fb/utils.py

The targetcli utility checks WWPN when adding initiators in LIO.

Hyper-V virtual machines have a WWPN like "c003ff9bfee40008", "c003ff9bfee4000a" on Virtual FC HBA (connect to SAN via NPIV) ...

If we are trying to add Hyper-V VMs as the initiators in Debian 9 LIO via targetcli tool, we get the error “WWN not valid as: naa”

maurizio-lombardi commented 5 years ago

Thanks for the report, I am going to have a look and fix it.

maurizio-lombardi commented 5 years ago

naa': lambda wwn: re.match(r"naa.[125][0-9a-fA-F]{15}$", wwn), In my opinion, this line should be changed as follows: 'naa':lambda wwn: re.match("naa\.[0-9a-fA-F]{16}$",wwn),

Hmmm, this would accept everything and probably is not the correct thing to do. For example: if the first nibble is 0 the name is to be considered invalid, 0x7 - 0xb are reserved and we should print an error. Probably the correct thing to do is to introduce support for "EUI-64 Mapped" ( 0xc to 0xf).

In other words, something like this (not tested):

'naa': lambda wwn: re.match(r"naa\.[125c-fC-F][0-9a-fA-F]{15}$", wwn)

friedkiwi commented 2 years ago

This might be a really old issue, but I've recently ran into the same issue when trying to export LUNs to my IBM mainframe. The WWN it uses is c05076d1c4001001.

'naa': lambda wwn: re.match(r"naa.[125c-fC-F][0-9a-fA-F]{15}$", wwn)

@maurizio-lombardi I can confirm that this works

maurizio-lombardi commented 2 years ago

@friedkiwi the reason I didn't push the change back then was because no one tested if it works or introduce any regression. Now that you are able to test this we can move this thing forward. If you want to submit a pull request just go for it