unable to add IPv6-only interface address as iscsi portal

Bug #1641339 reported by Kevin Otte
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
rtslib (Ubuntu)
New
Undecided
Unassigned

Bug Description

When I attempt to add a ULA address as a portal IP:

/iscsi/iqn.20.../tpg1/portals> pwd
/iscsi/iqn.2000-12.net.nivex:storage.omashu/tpg1/portals
/iscsi/iqn.20.../tpg1/portals> create fd60:e0:a0f4:121::4
Using default IP port 3260
IP address does not exist: fd60:e0:a0f4:121::4

But this succeeds when adding a global IP:

/iscsi/iqn.20.../tpg1/portals> create 2606:a000:a449:5900::4
Using default IP port 3260
Created network portal 2606:a000:a449:5900::4:3260.
Entering new node /iscsi/iqn.2000-12.net.nivex:storage.omashu/tpg1/portals/2606:a000:a449:5900::4:3260

I do not know if this is a filter within the target that it can only add global IPs, or if it is only looking at primary interfaces and not VLANs:

$ ip addr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
    link/ether 00:21:85:10:77:49 brd ff:ff:ff:ff:ff:ff
    inet 172.31.3.4/24 brd 172.31.3.255 scope global eth0
       valid_lft forever preferred_lft forever
    inet6 2606:a000:a449:5900::4/64 scope global
       valid_lft forever preferred_lft forever
    inet6 fe80::221:85ff:fe10:7749/64 scope link
       valid_lft forever preferred_lft forever
3: eth0.121@eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
    link/ether 00:21:85:10:77:49 brd ff:ff:ff:ff:ff:ff
    inet6 fd60:e0:a0f4:121::4/64 scope global
       valid_lft forever preferred_lft forever
    inet6 fe80::221:85ff:fe10:7749/64 scope link
       valid_lft forever preferred_lft forever

ProblemType: Bug
DistroRelease: Ubuntu 16.04
Package: targetcli 1:3.0~pre4.1~ga55d018-2
ProcVersionSignature: Ubuntu 4.4.0-43.63-generic 4.4.21
Uname: Linux 4.4.0-43-generic x86_64
ApportVersion: 2.20.1-0ubuntu2.1
Architecture: amd64
Date: Sat Nov 12 13:55:57 2016
PackageArchitecture: all
ProcEnviron:
 TERM=xterm
 PATH=(custom, no user)
 XDG_RUNTIME_DIR=<set>
 LANG=en_US.UTF-8
 SHELL=/bin/bash
SourcePackage: targetcli
UpgradeStatus: Upgraded to xenial on 2016-09-05 (68 days ago)

Revision history for this message
Kevin Otte (nivex) wrote :
Revision history for this message
Kevin Otte (nivex) wrote :

Ah, this appears to be a fault in rtslib not enumerating the VLAN interface:

>>> import rtslib.utils
>>> rtslib.utils.list_eth_names()
['lo', 'eth0']

affects: targetcli (Ubuntu) → rtslib (Ubuntu)
summary: - unable to add ULA address as iscsi portal
+ unable to add VLAN interface address as iscsi portal
Revision history for this message
Kevin Otte (nivex) wrote :

So it turns out this is a bit of a "perfect storm" involving an IPv6-only interface.
I've filed a bug upstream at: https://github.com/Datera/rtslib/issues/17

summary: - unable to add VLAN interface address as iscsi portal
+ unable to add IPv6-only interface address as iscsi portal
Revision history for this message
Kevin Otte (nivex) wrote :

There has been no activity on this bug in nearly a year. We're a week away from releasing 17.10, which will put us well on the way to being in the dev cycle for the next LTS. I have grave concerns about a fix making it into the repo in time having witnessed deadlines come and go on previous releases. I really don't want to have to keep local patches around post-upgrade next year.

The upstream bug does have a patch associated with it. Is there anything else I need to do to shepherd this along?

Revision history for this message
Kevin Otte (nivex) wrote :

I just tested this with targetcli-fb in 17.10 and it gleefully allowed me to add a portal for an address that isn't even on the system, but I do see it listening in "ss -a -6" output. I guess this at least avoids the problem of upgrading my server from 16.04 -> 18.04 when the time comes. Sadly it looks as though 16.04 will be forever broken.

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.