Jafner/Jafner.net
1
0
Fork 0
Code Issues 16 Pull Requests Actions Packages Projects Releases Wiki Activity
fb44b723f3
Jafner.net/homelab/docs/iSCSI.md
Joey Hafner 6086222503
Reorganize root level of repo. 
- Move homelab, Jafner.dev (now called blog) to root.
- Rename "archived projects" -> "archive"
- Rename "active projects" -> "projects"
- Rename "jafner-homebrew" -> "5ehomebrew"
- Rename "docker-llm-amd" -> "local-ai"
2024-07-16 12:17:55 -07:00

9.2 KiB
Raw Blame History

NOTE: Adding or removing drives

The drive letter of the iSCSI device will change (e.g. from /dev/sde to /dev/sdb) if drives are added or removed. This will cause the mount to fail.

To resolve: 0. Make sure all Docker stacks relying on the iSCSI drive are shut down.

  1. Update the fstab entry. Edit the /etc/fstab file as root, and update the drive letter.
  2. Re-mount the drive. Run sudo mount -a.

Creating the Zvol and iSCSI share in TrueNAS Scale

  1. Navigate to the dataset to use. From the TrueNAS Scale dashboard, open the navigation side panel. Navigate to "Datasets". Select the pool to use (Tank).
  2. Create the Zvol to use. In the top-left, click "Add Zvol" (Why not a dataset?). Name: fighter, Size for this zvol: 8 TiB. Leave all other settings default.
  3. Navigate to the iSCSI share creator. Navigate to "Shares". Open the "Block (iSCSI) Shares Targets" panel. (Optionally, set the base name per RFC 3721 1.1 (iqn.2020-04.net.jafner)).
  4. Create the iSCSI share. Click the "Wizard" button in the top-right. a. Create or Choose Block Device. Name: fighter, Device: zvol/Tank/fighter, Sharing Platform: Modern OS. b. Portal. Portal: Create New, Discovery Authentication Method: NONE, Discovery Authentication Group: NONE, Add listen: 0.0.0.0. c. Initiator. Leave blank to allow all hostnames and IPs to initiate. Optionally enter a list IP address(es) or hostname(s) to restrict access to the device. d. Confirm. Review and Save.
  5. Enable iSCSI service at startup. Navigate to System Settings -> Services. If it's not already running, enable the iSCSI service and check the box to "Start Automatically".

Connecting to the iSCSI target

  1. Install the open-iscsi package.
  • sudo apt-get install open-iscsi
  1. Get the list of available shares.
    • sudo iscsiadm --mode discovery --type sendtargets --portal 192.168.1.10
    • The IP for --portal is the IP of the NAS hosting the iSCSI share.
    • In my case, this command returns 192.168.1.10:3260,1 iqn.2020-03.net.jafner:fighter.
  2. Open the iSCSI session.
    • sudo iscsiadm --mode node --targetname "iqn.2020-03.net.jafner:fighter" --portal "192.168.1.10:3260" --login
    • The name for --targetname is the iqn string including the share name.
    • The address for --portal has both the IP and port used by the NAS hosting the iSCSI share.
  3. Verify the session connected.
    • sudo iscsiadm --mode session --print=1
    • This should return the description of any active sessions.

Debian.org.

Initializing the iSCSI disk

  1. Identify the device name of the new device with sudo iscsiadm -m session -P 3 | grep "Attached scsi disk". In my case, sdb. ServerFault.
  2. Partition and format the device. Run sudo parted --script /dev/sdb "mklabel gpt" && sudo parted --script /dev/sdb "mkpart primary 0% 100%" && sudo mkfs.ext4 /dev/sdb1 Server-world.info.
  3. Mount the new partition to a directory. Run sudo mkdir /mnt/iscsi && sudo mount /dev/sdb1 /mnt/iscsi. Where the path /dev/sdb1 is the newly-created partition and the path /mnt/iscsi is the path to which you want it mounted.
  4. Test the disk write speed of the new partition. Run sudo dd if=/dev/zero of=/mnt/iscsi/temp.tmp bs=1M count=32768 to run a 32GB test write. Cloudzy.com.

Connecting and mounting the iSCSI share on boot

  1. Get the full path of the share's configuration. It should be like /etc/iscsi/nodes/<share iqn>/<share host address>/default. In my case it was /etc/iscsi/nodes/iqn.2020-03.net.jafner:fighter/192.168.1.10,3260,1/default. Debian.org.
  2. Set the node.startup parameter to automatic. Run sudo sed -i 's/node.startup = manual/node.startup = automatic/g' /etc/iscsi/nodes/iqn.2020-03.net.jafner:fighter/192.168.1.10,3260,1/default.
  3. Add the new mount to /etc/fstab. Run sudo bash -c "echo '/dev/sdb1 /mnt/iscsi ext4 _netdev 0 0' >> /etc/fstab". Adamsdesk.com, StackExchange.

How to Gracefully Terminate iSCSI Session

  1. Ensure any Docker containers currently using the device are shut down. Run for stack in /home/admin/homelab/fighter/config/*; do cd $stack && if $(docker-compose config | grep -q /mnt/iscsi); then echo "ISCSI-DEPENDENT: $stack"; fi ; done to get the list of iSCSI-dependent stacks. Ensure all listed stacks are OK to shut down, then run for stack in /home/admin/homelab/fighter/config/*; do cd $stack && if $(docker-compose config | grep -q /mnt/iscsi); then echo "SHUTTING DOWN $stack" && docker-compose down; fi ; done.
  2. Unmount the iSCSI device. Run sudo umount /mnt/iscsi.
  3. Log out of the iSCSI session. Run sudo iscsiadm --mode node --targetname "iqn.2020-03.net.jafner:fighter" --portal "192.168.1.10:3260" --logout.
  4. Shut down the host. Run sudo shutdown now.

Systemd-ifying the process

Remove the iSCSI mount from /etc/fstab, but otherwise most of the steps above should be fine. (Don't forget to install and enable the iscsid.service systemd unit).

Script for connecting to (and disconnecting from) iSCSI session

This script is one command, but sometimes it's useful to contain it in a script. connect-iscsi.sh

#!/bin/bash
iscsiadm --mode node --targetname iqn.2020-03.net.jafner:fighter --portal 192.168.1.10:3260 --login

disconnect-iscsi.sh

#!/bin/bash
iscsiadm --mode node --targetname iqn.2020-03.net.jafner:fighter --portal 192.168.1.10:3260, 1 -u

Systemd Unit for connecting iSCSI session

/etc/systemd/system/connect-iscsi.service with root:root 644 permissions

[Unit]
Description=Connect iSCSI session
Requires=network-online.target
#After=
DefaultDependencies=no

[Service]
User=root
Group=root
Type=oneshot
RemainAfterExit=true
ExecStart=iscsiadm --mode node --targetname iqn.2020-03.net.jafner:fighter --portal 192.168.1.10:3260 --login
StandardOutput=journal

[Install]
WantedBy=multi-user.target

Systemd Unit for mounting the share

/etc/systemd/system/mnt-nas-iscsi.mount with root:root 644 permissions Note that the file name must be mnt-nas-iscsi if its Where= parameter is /mnt/nas/iscsi. Docs

[Unit]
Description="Mount iSCSI share /mnt/nas/iscsi"
After=connect-iscsi.service
DefaultDependencies=no

[Mount]
What=/dev/disk/by-uuid/cf3a253c-e792-48b5-89a1-f91deb02b3be
Where=/mnt/nas/iscsi
Type=ext4
StandardOutput=journal

[Install]
WantedBy=multi-user.target

Systemd Unit for automounting the share

/etc/systemd/system/mnt-nas-iscsi.automount with root:root 644 permissions Note that the file name must be mnt-nas-iscsi if its Where= parameter is /mnt/nas/iscsi. Docs

[Unit]
Description="Mount iSCSI share /mnt/nas/iscsi"
Requires=network-online.target
#After=

[Automount]
Where=/mnt/nas/iscsi

[Install]
WantedBy=multi-user.target

Quick interactive one-liner to install these scripts

This will open each file for editing in nano under the path /etc/systemd/system/ and apply the correct permissions to the file after it has been written.

for file in /etc/systemd/system/connect-iscsi.service /etc/systemd/system/mnt-nas-iscsi.mount /etc/systemd/system/mnt-nas-iscsi.automount; do sudo nano $file && sudo chown root:root $file && sudo chmod 644 $file && sudo systemctl enable $(basename $file); done && sudo systemctl daemon-reload

After this, it's probably a good idea to reboot from scratch.

Check statuses

  • sudo systemctl status connect-iscsi.service
  • sudo systemctl status mnt-nas-iscsi.mount
  • sudo systemctl status mnt-nas-iscsi.automount

https://unix.stackexchange.com/questions/195116/mount-iscsi-drive-at-boot-system-halts https://github.com/f1linux/iscsi-automount/blob/master/config-iscsi-storage.sh https://github.com/f1linux/iscsi-automount/blob/master/config-iscsi-storage-mounts.sh

Disabling all iSCSI units for debugging

During an extended outage of barbarian, we learned that, as configured, fighter will not boot while its iSCSI target is inaccessible. To resolve, we disabled the following systemd units:

iscsi.service
mnt-nas-iscsi.automount
mnt-nas-iscsi.mount
connect-iscsi.service
barbarian-wait-online.service
iscsid.service

Oneliners below:

  • Disable: for unit in iscsi.service mnt-nas-iscsi.automount mnt-nas-iscsi.mount connect-iscsi.service barbarian-wait-online.service iscsid.service; do systemctl disable $unit; done
  • Enable: for unit in iscsi.service mnt-nas-iscsi.automount mnt-nas-iscsi.mount connect-iscsi.service barbarian-wait-online.service iscsid.service; do systemctl enable $unit; done