open-iscsi does not automatically login into portal about open-iscsi

[gonzoleeman]

I know the automatic login stuff works, so let's figure out why it's not working here.

First of all, which distribution is this? Where did you get the open-iscsi package you are using?

Next, it seems like you are mixing old-school and new-school methods for starting your service. Newer open-iscsi packages use systemd, which means (1) they do not use /etc/init.d scripts, and (2) you should never set the "iscsid.startup = /some/path/to/iscsid". That is because systemd does all the starting/stopping/restarting for you.

Assuming you are using systemd, what systemd services to you have for iscsi? I see you have ran "systemctl start open-iscsi.service". Are there any others? (Run "ls /usr/lib/systemd/system/iscsi" as root.) I'm trying to understand if your open-iscsi is socket-activated or not.

Lastly, just as a guess, I never say that you ran "systemctl enable open-iscsi.service". This tells sytemd to start your service at boot time, whereas "systemctl start open-iscsi.service" just starts it up right now.

[myghacc]

Distribution is Mageia 6 64-bit with open-iscsi from their repositories.

I deleted everything in /var/lib/iscsi and uninstalled open-iscsi. After that I created a new interface with iscsiadm (iscsiadm -m iface -I icebox --op=new) and re-installed open-iscsi. Then I did a discovery with iscsiadm (iscsiadm --mode discovery --type sendtargets --portal 192.168.10.10 --discover) and checked that the portal I want to use is set to automatic (cat nodes/iqn.2010-08.eu.n-tec:icebox-rix6r-000905600:dev0.ctr1/192.168.10.10,3260,0/icebox | grep automatic). Finally I rebooted the system. The following output I get after the reboot:

[root@thalamus_nfs iscsi]# systemctl status open-iscsi.service
● open-iscsi.service - Open iSCSI Daemon
Loaded: loaded (/usr/lib/systemd/system/open-iscsi.service; enabled; vendor preset: enabled)
Active: active (running) since Di 2017-12-19 08:25:06 CET; 5min ago
Process: 2868 ExecStart=/sbin/iscsid (code=exited, status=0/SUCCESS)
Process: 2808 ExecStartPre=/usr/sbin/modprobe -q iscsi_tcp (code=exited, status=0/SUCCESS)
CGroup: /system.slice/open-iscsi.service
├─2872 /sbin/iscsid
└─2873 /sbin/iscsid

Dez 19 08:25:06 thalamus_nfs systemd[1]: Starting Open iSCSI Daemon...
Dez 19 08:25:06 thalamus_nfs systemd[1]: Started Open iSCSI Daemon.
Dez 19 08:25:07 thalamus_nfs iscsid[2872]: transport class version 2.0-870. iscsid version 2.0-872
Dez 19 08:25:07 thalamus_nfs iscsid[2872]: iSCSI daemon with pid=2873 started!
[root@thalamus_nfs iscsi]# pwd
/var/lib/iscsi
[root@thalamus_nfs iscsi]# tree
.
├── ifaces
│   └── icebox
├── nodes
│   └── iqn.2010-08.eu.n-tec:icebox-rix6r-000905600:dev0.ctr1
│   ├── 192.168.10.10,3260,0
│   │   └── icebox
│   ├── 192.168.20.10,3260,0
│   │   └── icebox
│   ├── 192.168.3.1,3260,1
│   │   └── icebox
│   └── 192.168.4.1,3260,1
│   └── icebox
└── send_targets
└── icebox_iscsi_1,3260
├── iqn.2010-08.eu.n-tec:icebox-rix6r-000905600:dev0.ctr1,192.168.10.10,3260,0,icebox -> /var/lib/iscsi/nodes/iqn.2010-08.eu.n-tec:icebox-rix6r-000905600:dev0.ctr1/192.168.10.10,3260,0
├── iqn.2010-08.eu.n-tec:icebox-rix6r-000905600:dev0.ctr1,192.168.20.10,3260,0,icebox -> /var/lib/iscsi/nodes/iqn.2010-08.eu.n-tec:icebox-rix6r-000905600:dev0.ctr1/192.168.20.10,3260,0
├── iqn.2010-08.eu.n-tec:icebox-rix6r-000905600:dev0.ctr1,192.168.3.1,3260,1,icebox -> /var/lib/iscsi/nodes/iqn.2010-08.eu.n-tec:icebox-rix6r-000905600:dev0.ctr1/192.168.3.1,3260,1
├── iqn.2010-08.eu.n-tec:icebox-rix6r-000905600:dev0.ctr1,192.168.4.1,3260,1,icebox -> /var/lib/iscsi/nodes/iqn.2010-08.eu.n-tec:icebox-rix6r-000905600:dev0.ctr1/192.168.4.1,3260,1
└── st_config

13 directories, 6 files
[root@thalamus_nfs iscsi]# ls /usr/lib/systemd/system/iscsi
ls: Zugriff auf '/usr/lib/systemd/system/iscsi' nicht möglich: Datei oder Verzeichnis nicht gefunden
[root@thalamus_nfs iscsi]#

I can somehow solve this problem with a self-written systemd service. But still I am wondering why the missing step is not done automatically although open-iscsi daemon is running.

login-iscsi.service:

[Unit]
Description=Login into ISCSI portal
Requires=network.target open-iscsi.service
After=network.target open-iscsi.service

[Service]
Type=oneshot
RemainAfterExit=yes

ExecStart=/usr/sbin/iscsiadm --mode node --targetname iqn.2010-08.eu.n-tec:icebox-rix6r-000905600:dev0.ctr1 --portal 192.168.10.10:3260 --login
ExecStartPost=/usr/sbin/vgchange -a y

ExecStop=/usr/sbin/iscsiadm --mode node --targetname iqn.2010-08.eu.n-tec:icebox-rix6r-000905600:dev0.ctr1 --portal 192.168.10.10:3260 --logout

[Install]
WantedBy=multi-user.target

[gonzoleeman]

I think I understand your issue. You need an "iscsi login/logout" service. In the distributions that I've looked at, there are generally 3 systemd unit files for iscsi:

1. a "socket" file (e.g. "iscsid.socket"), which allows "socket activation" of the next service if somebody tries to talk to it
2. an "iscsid service" file (e.g. "iscsid.service"), which controls the starting and stopping of the iscsid daemon, and
3. an "iscsi login" file (e.g. "iscsi.service"), which controls starting and stopping "sessions" (i.e. logging into and out of targets)

The example names I gave are from openSUSE.

It is unfortunate that these files are not part of the open-iscsi distribution, but the problem is that these sets of services are implemented differently for each distribution, so their only value to others is as an example. And there's not much motivation to keep examples up to date.

Here is the "iscsi.service" file for openSUSE Leap 42.2:

[Unit]
Description=Login and scanning of iSCSI devices
Documentation=man:iscsiadm(8) man:iscsid(8)
After=network.target network-online.target iscsid.service
ConditionPathExists=/etc/iscsi/initiatorname.iscsi

[Service]
Type=oneshot
ExecStart=-/sbin/iscsiadm -m node --loginall=automatic
ExecStop=/sbin/iscsiadm -m node --logoutall=automatic
ExecStop=/sbin/iscsiadm -m node --logoutall=manual
SuccessExitStatus=21
RemainAfterExit=true

[Install]
WantedBy=remote-fs.target

[myghacc]

Thank you. I informed the Mageia bug team about this issue. Since it is working for me now I will close this issue.