qa/cephadm: extend iscsi teuth test

This extends the basic cephadm iscsi deployment test
to do some very basic gw and client setup. The iscsi
service setup is likely to be altered with the future
addition of sidecar containers in cephadm, so the
hope is this will provide a bit more coverage

Signed-off-by: Adam King <adking@redhat.com>

diff --git a/qa/suites/orch/cephadm/workunits/task/test_iscsi_container/test_iscsi_container.yaml b/qa/suites/orch/cephadm/workunits/task/test_iscsi_container/test_iscsi_container.yaml
index 19d302c87deb734522ebdeaacb681fef308a6aea..909b5ef0fc35552da92d63889e76eb8036652b10 100644 (file)
--- a/qa/suites/orch/cephadm/workunits/task/test_iscsi_container/test_iscsi_container.yaml
+++ b/qa/suites/orch/cephadm/workunits/task/test_iscsi_container/test_iscsi_container.yaml
@@ -19,3 +19,4 @@ tasks:
       client.0:
         - cephadm/test_iscsi_pids_limit.sh
         - cephadm/test_iscsi_etc_hosts.sh
+        - cephadm/test_iscsi_setup.sh

diff --git a/qa/workunits/cephadm/test_iscsi_setup.sh b/qa/workunits/cephadm/test_iscsi_setup.sh
new file mode 100755 (executable)
index 0000000..88f3799
--- /dev/null
+++ b/qa/workunits/cephadm/test_iscsi_setup.sh
@@ -0,0 +1,99 @@
+#!/bin/bash
+
+# very basic set up of iscsi gw and client
+# to make sure things are working
+
+set -ex
+
+if ! grep -q rhel /etc/*-release; then
+    echo "The script only supports CentOS."
+    exit 1
+fi
+
+# teuthology tends to put the cephadm binary built for our testing
+# branch in /home/ubuntu/cephtest/. If it's there, lets just move it
+# so we don't need to reference the full path.
+if ! command -v cephadm && ls /home/ubuntu/cephtest/cephadm; then
+    sudo cp /home/ubuntu/cephtest/cephadm /usr/sbin/
+fi
+
+# make sure we haven't already created luns
+! sudo ls /dev/disk/by-path | grep iscsi
+
+sudo dnf install jq -y
+
+ISCSI_CONT_ID=$(sudo podman ps -qa --filter='name=iscsi' | head -n 1)
+ISCSI_DAEMON_NAME=$(sudo cephadm ls --no-detail | jq -r '.[] | select(.name | startswith("iscsi")) | .name')
+ISCSI_DAEMON_ID=$(cut -d '.' -f2- <<< "$ISCSI_DAEMON_NAME")
+HOSTNAME=$(sudo cephadm shell -- ceph orch ps --daemon-id "$ISCSI_DAEMON_ID" -f json | jq -r '.[] | .hostname')
+NODE_IP=$(sudo cephadm shell -- ceph orch host ls --format json | jq --arg HOSTNAME "$HOSTNAME" -r '.[] | select(.hostname == $HOSTNAME) | .addr')
+# The result of this python line is what iscsi will expect for the first gateway name
+FQDN=$(python3 -c 'import socket; print(socket.getfqdn())')
+# I am running this twice on purpose. I don't know why but in my testing the first time this would
+# run it would return a different result then all subsequent runs (and take significantly longer to run).
+# The result from the first run would cause gateway creation to fail when the return value is used
+# later on. It was likely specific to my env, but it doesn't hurt to run it twice anyway. This
+# was the case whether I ran it through cephadm shell or directly on the host machine.
+FQDN=$(python3 -c 'import socket; print(socket.getfqdn())')
+ISCSI_POOL=$(sudo cephadm shell -- ceph orch ls iscsi --format json | jq -r '.[] | .spec | .pool')
+ISCSI_USER="adminadmin"
+ISCSI_PASSWORD="adminadminadmin"
+
+# gateway setup
+container_gwcli() {
+    sudo podman exec -it ${ISCSI_CONT_ID} gwcli "$@"
+}
+
+container_gwcli /iscsi-targets create iqn.2003-01.com.redhat.iscsi-gw:iscsi-igw
+# I've seen this give a nonzero error code with an error message even when
+# creating the gateway successfully, so this command is allowed to fail
+# If it actually failed to make the gateway, some of the follow up commands will fail
+container_gwcli /iscsi-targets/iqn.2003-01.com.redhat.iscsi-gw:iscsi-igw/gateways create ${FQDN} ${NODE_IP} || true
+container_gwcli /disks create pool=${ISCSI_POOL} image=disk_1 size=2G
+container_gwcli /disks create pool=${ISCSI_POOL} image=disk_2 size=2G
+container_gwcli /iscsi-targets/iqn.2003-01.com.redhat.iscsi-gw:iscsi-igw/hosts create iqn.1994-05.com.redhat:client1
+container_gwcli /iscsi-targets/iqn.2003-01.com.redhat.iscsi-gw:iscsi-igw/hosts/iqn.1994-05.com.redhat:client1 auth username=${ISCSI_USER}  password=${ISCSI_PASSWORD}
+container_gwcli /iscsi-targets/iqn.2003-01.com.redhat.iscsi-gw:iscsi-igw/hosts/iqn.1994-05.com.redhat:client1 disk add ${ISCSI_POOL}/disk_1
+container_gwcli /iscsi-targets/iqn.2003-01.com.redhat.iscsi-gw:iscsi-igw/hosts/iqn.1994-05.com.redhat:client1 disk add ${ISCSI_POOL}/disk_2
+
+# set up multipath and some iscsi config options
+sudo dnf install -y iscsi-initiator-utils device-mapper-multipath
+
+# this next line is purposely being done without "-a" on the tee command to
+# overwrite the current initiatorname.iscsi file if it is there
+echo "GenerateName=no" | sudo tee /etc/iscsi/initiatorname.iscsi
+echo "InitiatorName=iqn.1994-05.com.redhat:client1" | sudo tee -a /etc/iscsi/initiatorname.iscsi
+
+echo "node.session.auth.authmethod = CHAP" | sudo tee -a /etc/iscsi/iscsid.conf
+echo "node.session.auth.username = ${ISCSI_USER}" | sudo tee -a /etc/iscsi/iscsid.conf
+echo "node.session.auth.password = ${ISCSI_PASSWORD}" | sudo tee -a /etc/iscsi/iscsid.conf
+
+sudo tee -a /etc/multipath.conf > /dev/null << EOF
+devices {
+  device {
+          vendor                 "LIO-ORG"
+          product                "TCMU device"
+          hardware_handler       "1 alua"
+          path_grouping_policy   "failover"
+          path_selector          "queue-length 0"
+          failback               60
+          path_checker           tur
+          prio                   alua
+          prio_args              exclusive_pref_bit
+          fast_io_fail_tmo       25
+          no_path_retry          queue
+  }
+}
+EOF
+sudo systemctl restart multipathd
+sudo systemctl restart iscsid
+
+# client setup
+sudo iscsiadm -m discovery -t st -p ${NODE_IP}
+sudo iscsiadm -m node -T iqn.2003-01.com.redhat.iscsi-gw:iscsi-igw -l
+sudo iscsiadm -m session --rescan
+
+sleep 5
+
+# make sure we can now see luns
+sudo ls /dev/disk/by-path | grep iscsi