FreeBSD rc.script for activating ZFS pool over iSCSI

iscsi-zfs

#!/bin/sh

# PROVIDE: iscsi-zfs
# REQUIRE: iscsid iscsictl mountcritlocal var

. /etc/rc.subr

name="iscsi_zfs"
start_cmd="iscsi_zfs_start"
stop_cmd="iscsi_zfs_stop"
rcvar="iscsi_zfs_enable"

iscsi_zfs_enable=${iscsi_zfs_enable:-"NO"}
iscsi_zfs_wait_iqn=${iscsi_zfs_wait_iqn:-""}
iscsi_zfs_wait_pool=${iscsi_zfs_wait_pool:-""}
iscsi_zfs_wait_retry=${iscsi_zfs_wait_retry:-"30"}

iscsi_zfs_start()
{
    retry=0
    while [ "$retry" -lt "${iscsi_zfs_wait_retry}" ]
    do
        if iscsictl | fgrep "${iscsi_zfs_wait_iqn}" | fgrep "Connected: " >/dev/null; then
            break
        fi
        retry=`expr ${retry} + 1`
        sleep 1
    done
    while [ "$retry" -lt "${iscsi_zfs_wait_retry}" ]
    do
        if zpool status "${iscsi_zfs_wait_pool}" >/dev/null 2>&1; then
            break
        fi
        retry=`expr ${retry} + 1`
        sleep 1
    done
    while [ "$retry" -lt "${iscsi_zfs_wait_retry}" ]
    do
        zfs mount -a
        if mount | fgrep "${iscsi_zfs_wait_pool}" >/dev/null 2>&1; then
            break
        fi
        retry=`expr ${retry} + 1`
        sleep 1
    done
}

load_rc_config $name
run_rc_command "$1"

Hi

Thanks for your script. I have an issue I can not resolve. Script does not work for some reason. I appreciate your help

root@iscsi:/usr/local/etc/rc.d # sh ./iscsi_zfs start
export: -zfs_env: bad variable name
root@iscsi:/usr/local/etc/rc.d # uname -a
FreeBSD iscsi 13.0-RELEASE FreeBSD 13.0-RELEASE #0 releng/13.0-n244733-ea31abc261f: Fri Apr 9 04:24:09 UTC 2021 root@releng1.nyi.freebsd.org:/usr/obj/usr/src/amd64.amd64/sys/GENERIC amd64
root@iscsi:/usr/local/etc/rc.d #
root@iscsi:/usr/local/etc/rc.d # iscsictl
Target name Target portal State
iqn.2005-10.org.freenas.ctl:test 192.168.4.7 Connected: da1
root@iscsi:/usr/local/etc/rc.d # zpool status
no pools available
root@iscsi:/usr/local/etc/rc.d #

@amikhail74

name="iscsi-zfs"

This line is causing the problem. It seems name should not contain hyphens nowadays.

Also, a dependency (BEFORE) to xtreemfs-osd, that is useless unless you are using xtreemfs, can be removed.

Thanks a lot for your help! Now it starts but it does not seem to work. The problem that the script expects pools already be imported. But the fact is that after reboot the list of pools is empty. It requres zpool import to be done first. Maybe either the issue with the local iscsictl configuration, or remote iscsi share, or zfs pool requires some special setting.

Upd: It seems that there is a difference between FreeBSD 13 and FreeBSD 12.2. 12.2 works as expected. After reboot 12.2 has a zfs pool in place

Yes, this script expects target pools to be imported already (and imported state persists over reboots).

I'm not sure whether imported pools persists over reboot, but adding zpool import "${iscsi_zfs_wait_pool}" in retry-loop may help.

     while [ "$retry" -lt "${iscsi_zfs_wait_retry}" ]
     do
         if zpool status "${iscsi_zfs_wait_pool}" >/dev/null 2>&1; then
             break
         fi
+        zpool import "${iscsi_zfs_wait_pool}"
         retry=`expr ${retry} + 1`
         sleep 1

It can be even more simple. 'zpool import -a' mounts all imported pools.

Thanks!