Bump version to 0.9.93

Even if the subfunction of an API @ZKConnection call fails, the
zkhandler needs to terminate and clean up, or it leaves stuck threads
around.

.version

@@ -1 +1 @@
-0.9.90
+0.9.93

CHANGELOG.md

@@ -1,5 +1,25 @@
 ## PVC Changelog
 
+###### [v0.9.93](https://github.com/parallelvirtualcluster/pvc/releases/tag/v0.9.93)
+
+  * [API Daemon] Fixes a bug where stuck zkhandler threads were not cleaned up on error
+
+###### [v0.9.92](https://github.com/parallelvirtualcluster/pvc/releases/tag/v0.9.92)
+
+  * [CLI Client] Adds the new restore state to the colours list for VM status
+  * [API Daemon] Fixes an incorrect variable assignment
+  * [Provisioner] Improves the error handling of various steps in the debootstrap and rinse example scripts
+  * [CLI Client] Fixes two bugs around missing keys that were added recently (uses get() instead direct dictionary refs)
+  * [CLI Client] Improves API error handling via GET retries (x3) and better server status code handling
+
+###### [v0.9.91](https://github.com/parallelvirtualcluster/pvc/releases/tag/v0.9.91)
+
+  * [Client CLI] Fixes a bug and improves output during cluster task events.
+  * [Client CLI] Improves the output of the task list display.
+  * [Provisioner] Fixes some missing cloud-init modules in the default debootstrap script.
+  * [Client CLI] Fixes a bug with a missing argument to the vm_define helper function.
+  * [All] Fixes inconsistent package find + rm commands to avoid errors in dpkg.
+
 ###### [v0.9.90](https://github.com/parallelvirtualcluster/pvc/releases/tag/v0.9.90)
 
   * [Client CLI/API Daemon] Adds additional backup metainfo and an emailed report option to autobackups.

api-daemon/provisioner/examples/script/3-debootstrap.py

@@ -150,6 +150,10 @@
 from daemon_lib.vmbuilder import VMBuilder
 
 
+# These are some global variables used below
+default_root_password = "test123"
+
+
 # The VMBuilderScript class must be named as such, and extend VMBuilder.
 class VMBuilderScript(VMBuilder):
     def setup(self):
@@ -498,11 +502,15 @@ class VMBuilderScript(VMBuilder):
         ret = os.system(
             f"debootstrap --include={','.join(deb_packages)} {deb_release} {temp_dir} {deb_mirror}"
         )
+        ret = int(ret >> 8)
         if ret > 0:
-            self.fail("Failed to run debootstrap")
+            self.fail(f"Debootstrap failed with exit code {ret}")
 
         # Bind mount the devfs so we can grub-install later
-        os.system("mount --bind /dev {}/dev".format(temp_dir))
+        ret = os.system("mount --bind /dev {}/dev".format(temp_dir))
+        ret = int(ret >> 8)
+        if ret > 0:
+            self.fail(f"/dev bind mount failed with exit code {ret}")
 
         # Create an fstab entry for each volume
         fstab_file = "{}/etc/fstab".format(temp_dir)
@@ -688,23 +696,36 @@ GRUB_DISABLE_LINUX_UUID=false
         # Do some tasks inside the chroot using the provided context manager
         with chroot(temp_dir):
             # Install and update GRUB
-            os.system(
+            ret = os.system(
                 "grub-install --force /dev/rbd/{}/{}_{}".format(
                     root_volume["pool"], vm_name, root_volume["disk_id"]
                 )
             )
-            os.system("update-grub")
+            ret = int(ret >> 8)
+            if ret > 0:
+                self.fail(f"GRUB install failed with exit code {ret}")
+
+            ret = os.system("update-grub")
+            ret = int(ret >> 8)
+            if ret > 0:
+                self.fail(f"GRUB update failed with exit code {ret}")
 
             # Set a really dumb root password so the VM can be debugged
             # EITHER CHANGE THIS YOURSELF, here or in Userdata, or run something after install
             # to change the root password: don't leave it like this on an Internet-facing machine!
-            os.system("echo root:test123 | chpasswd")
+            ret = os.system(f"echo root:{default_root_password} | chpasswd")
+            ret = int(ret >> 8)
+            if ret > 0:
+                self.fail(f"Root password change failed with exit code {ret}")
 
             # Enable cloud-init target on (first) boot
             # Your user-data should handle this and disable it once done, or things get messy.
             # That cloud-init won't run without this hack seems like a bug... but even the official
             # Debian cloud images are affected, so who knows.
-            os.system("systemctl enable cloud-init.target")
+            ret = os.system("systemctl enable cloud-init.target")
+            ret = int(ret >> 8)
+            if ret > 0:
+                self.fail(f"Enable of cloud-init failed with exit code {ret}")
 
     def cleanup(self):
         """
@@ -729,7 +750,7 @@ GRUB_DISABLE_LINUX_UUID=false
         temp_dir = "/tmp/target"
 
         # Unmount the bound devfs
-        os.system("umount {}/dev".format(temp_dir))
+        os.system("umount -f {}/dev".format(temp_dir))
 
         # Use this construct for reversing the list, as the normal reverse() messes with the list
         for volume in list(reversed(self.vm_data["volumes"])):
@@ -746,7 +767,7 @@ GRUB_DISABLE_LINUX_UUID=false
             ):
                 # Unmount filesystem
                 retcode, stdout, stderr = pvc_common.run_os_command(
-                    f"umount {mount_path}"
+                    f"umount -f {mount_path}"
                 )
                 if retcode:
                     self.log_err(

api-daemon/provisioner/examples/script/4-rinse.py

@@ -150,6 +150,11 @@
 from daemon_lib.vmbuilder import VMBuilder
 
 
+# These are some global variables used below
+default_root_password = "test123"
+default_local_time = "UTC"
+
+
 # The VMBuilderScript class must be named as such, and extend VMBuilder.
 class VMBuilderScript(VMBuilder):
     def setup(self):
@@ -524,13 +529,23 @@ class VMBuilderScript(VMBuilder):
         ret = os.system(
             f"rinse --arch {rinse_architecture} --directory {temporary_directory} --distribution {rinse_release} --cache-dir {rinse_cache} --add-pkg-list /tmp/addpkg --verbose {mirror_arg}"
         )
+        ret = int(ret >> 8)
         if ret > 0:
-            self.fail("Failed to run rinse")
+            self.fail(f"Rinse failed with exit code {ret}")
 
         # Bind mount the devfs, sysfs, and procfs so we can grub-install later
-        os.system("mount --bind /dev {}/dev".format(temporary_directory))
+        ret = os.system("mount --bind /dev {}/dev".format(temporary_directory))
-        os.system("mount --bind /sys {}/sys".format(temporary_directory))
+        ret = int(ret >> 8)
-        os.system("mount --bind /proc {}/proc".format(temporary_directory))
+        if ret > 0:
+            self.fail(f"/dev bind mount failed with exit code {ret}")
+        ret = os.system("mount --bind /sys {}/sys".format(temporary_directory))
+        ret = int(ret >> 8)
+        if ret > 0:
+            self.fail(f"/sys bind mount failed with exit code {ret}")
+        ret = os.system("mount --bind /proc {}/proc".format(temporary_directory))
+        ret = int(ret >> 8)
+        if ret > 0:
+            self.fail(f"/proc bind mount failed with exit code {ret}")
 
         # Create an fstab entry for each volume
         fstab_file = "{}/etc/fstab".format(temporary_directory)
@@ -642,41 +657,76 @@ GRUB_SERIAL_COMMAND="serial --speed=115200 --unit=0 --word=8 --parity=no --stop=
         # Do some tasks inside the chroot using the provided context manager
         with chroot(temporary_directory):
             # Fix the broken kernel from rinse by setting a systemd machine ID and running the post scripts
-            os.system("systemd-machine-id-setup")
+            ret = os.system("systemd-machine-id-setup")
-            os.system(
+            ret = int(ret >> 8)
+            if ret > 0:
+                self.fail(f"Machine ID setup failed with exit code {ret}")
+
+            ret = os.system(
                 "rpm -q --scripts kernel-core | grep -A20  'posttrans scriptlet' | tail -n+2 | bash -x"
             )
+            ret = int(ret >> 8)
+            if ret > 0:
+                self.fail(f"RPM kernel reinstall failed with exit code {ret}")
 
             # Install any post packages
-            os.system(f"dnf install -y {' '.join(post_packages)}")
+            if len(post_packages) > 0:
+                ret = os.system(f"dnf install -y {' '.join(post_packages)}")
+                ret = int(ret >> 8)
+                if ret > 0:
+                    self.fail(f"DNF install failed with exit code {ret}")
 
             # Install and update GRUB config
-            os.system(
+            ret = os.system(
                 "grub2-install --force /dev/rbd/{}/{}_{}".format(
                     root_volume["pool"], vm_name, root_volume["disk_id"]
                 )
             )
+            ret = int(ret >> 8)
+            if ret > 0:
+                self.fail(f"GRUB install failed with exit code {ret}")
+
             os.system("grub2-mkconfig -o /boot/grub2/grub.cfg")
+            ret = int(ret >> 8)
+            if ret > 0:
+                self.fail(f"GRUB update failed with exit code {ret}")
 
             # Set a really dumb root password so the VM can be debugged
             # EITHER CHANGE THIS YOURSELF, here or in Userdata, or run something after install
             # to change the root password: don't leave it like this on an Internet-facing machine!
-            os.system("echo root:test123 | chpasswd")
+            ret = os.system(f"echo root:{default_root_password} | chpasswd")
+            ret = int(ret >> 8)
+            if ret > 0:
+                self.fail(f"Root password change failed with exit code {ret}")
 
             # Enable dbus-broker
-            os.system("systemctl enable dbus-broker.service")
+            ret = os.system("systemctl enable dbus-broker.service")
+            ret = int(ret >> 8)
+            if ret > 0:
+                self.fail(f"Enable of dbus-broker failed with exit code {ret}")
 
             # Enable NetworkManager
             os.system("systemctl enable NetworkManager.service")
+            ret = int(ret >> 8)
+            if ret > 0:
+                self.fail(f"Enable of NetworkManager failed with exit code {ret}")
 
             # Enable cloud-init target on (first) boot
             # Your user-data should handle this and disable it once done, or things get messy.
             # That cloud-init won't run without this hack seems like a bug... but even the official
             # Debian cloud images are affected, so who knows.
             os.system("systemctl enable cloud-init.target")
+            ret = int(ret >> 8)
+            if ret > 0:
+                self.fail(f"Enable of cloud-init failed with exit code {ret}")
 
             # Set the timezone to UTC
-            os.system("ln -sf ../usr/share/zoneinfo/UTC /etc/localtime")
+            ret = os.system(
+                f"ln -sf ../usr/share/zoneinfo/{default_local_time} /etc/localtime"
+            )
+            ret = int(ret >> 8)
+            if ret > 0:
+                self.fail(f"Localtime update failed with exit code {ret}")
 
     def cleanup(self):
         """

api-daemon/pvcapid/Daemon.py

@@ -27,7 +27,7 @@ from distutils.util import strtobool as dustrtobool
 import daemon_lib.config as cfg
 
 # Daemon version
-version = "0.9.90"
+version = "0.9.93"
 
 # API version
 API_VERSION = 1.0

client-cli/pvc/lib/common.py

@@ -140,7 +140,12 @@ def call_api(
     # Determine the request type and hit the API
     disable_warnings()
     try:
+        response = None
         if operation == "get":
+            retry_on_code = [429, 500, 502, 503, 504]
+            for i in range(3):
+                failed = False
+                try:
                     response = requests.get(
                         uri,
                         timeout=timeout,
@@ -149,6 +154,17 @@ def call_api(
                         data=data,
                         verify=config["verify_ssl"],
                     )
+                    if response.status_code in retry_on_code:
+                        failed = True
+                        continue
+                except requests.exceptions.ConnectionError:
+                    failed = True
+                    pass
+            if failed:
+                error = f"Code {response.status_code}" if response else "Timeout"
+                raise requests.exceptions.ConnectionError(
+                    f"Failed to connect after 3 tries ({error})"
+                )
         if operation == "post":
             response = requests.post(
                 uri,
@@ -189,7 +205,8 @@ def call_api(
             )
     except Exception as e:
         message = "Failed to connect to the API: {}".format(e)
-        response = ErrorResponse({"message": message}, 500)
+        code = response.status_code if response else 504
+        response = ErrorResponse({"message": message}, code)
 
     # Display debug output
     if config["debug"]:

client-cli/pvc/lib/storage.py

@@ -430,7 +430,7 @@ def format_list_osd(config, osd_list):
             )
             continue
 
-        if osd_information["is_split"]:
+        if osd_information.get("is_split") is not None:
             osd_information["device"] = f"{osd_information['device']} [s]"
 
         # Deal with the size to human readable

client-cli/pvc/lib/vm.py

@@ -1632,6 +1632,7 @@ def format_info(config, domain_information, long_output):
         "migrate": ansiprint.blue(),
         "unmigrate": ansiprint.blue(),
         "provision": ansiprint.blue(),
+        "restore": ansiprint.blue(),
     }
     ainformation.append(
         "{}State:{}              {}{}{}".format(
@@ -1716,7 +1717,7 @@ def format_info(config, domain_information, long_output):
         "{}Max live downtime:{}  {}".format(
             ansiprint.purple(),
             ansiprint.end(),
-            f"{domain_information['migration_max_downtime']} ms",
+            f"{domain_information.get('migration_max_downtime')} ms",
         )
     )
 

client-cli/setup.py

@@ -2,7 +2,7 @@ from setuptools import setup
 
 setup(
     name="pvc",
-    version="0.9.90",
+    version="0.9.93",
     packages=["pvc.cli", "pvc.lib"],
     install_requires=[
         "Click",

daemon-common/cluster.py

@@ -1201,7 +1201,7 @@ def get_resource_metrics(zkhandler):
         try:
             user_time = vm["vcpu_stats"]["user_time"] / 1000000
         except Exception:
-            cpu_time = 0
+            user_time = 0
         output_lines.append(
             f"pvc_vm_vcpus_user_time{{vm=\"{vm['name']}\"}} {user_time}"
         )

daemon-common/zkhandler.py

@@ -57,8 +57,9 @@ class ZKConnection(object):
                 schema_version = 0
             zkhandler.schema.load(schema_version, quiet=True)
 
+            try:
                 ret = function(zkhandler, *args, **kwargs)
-
+            finally:
                 zkhandler.disconnect()
                 del zkhandler
 

debian/changelog

@@ -1,3 +1,29 @@
+pvc (0.9.93-0) unstable; urgency=high
+
+  * [API Daemon] Fixes a bug where stuck zkhandler threads were not cleaned up on error
+
+ -- Joshua M. Boniface <joshua@boniface.me>  Tue, 30 Jan 2024 09:51:21 -0500
+
+pvc (0.9.92-0) unstable; urgency=high
+
+  * [CLI Client] Adds the new restore state to the colours list for VM status
+  * [API Daemon] Fixes an incorrect variable assignment
+  * [Provisioner] Improves the error handling of various steps in the debootstrap and rinse example scripts
+  * [CLI Client] Fixes two bugs around missing keys that were added recently (uses get() instead direct dictionary refs)
+  * [CLI Client] Improves API error handling via GET retries (x3) and better server status code handling
+
+ -- Joshua M. Boniface <joshua@boniface.me>  Mon, 29 Jan 2024 09:39:10 -0500
+
+pvc (0.9.91-0) unstable; urgency=high
+
+  * [Client CLI] Fixes a bug and improves output during cluster task events.
+  * [Client CLI] Improves the output of the task list display.
+  * [Provisioner] Fixes some missing cloud-init modules in the default debootstrap script.
+  * [Client CLI] Fixes a bug with a missing argument to the vm_define helper function.
+  * [All] Fixes inconsistent package find + rm commands to avoid errors in dpkg.
+
+ -- Joshua M. Boniface <joshua@boniface.me>  Tue, 23 Jan 2024 10:02:19 -0500
+
 pvc (0.9.90-0) unstable; urgency=high
 
   * [Client CLI/API Daemon] Adds additional backup metainfo and an emailed report option to autobackups.

debian/pvc-client-cli.postinst

@@ -2,10 +2,12 @@
 
 # Generate the bash completion configuration
 if [ -d /etc/bash_completion.d ]; then
+    echo "Installing BASH completion configuration"
     _PVC_COMPLETE=source_bash pvc > /etc/bash_completion.d/pvc
 fi
 
-# Remove pycaches
+# Remove any cached CPython directories or files
-find /usr/lib/python3/dist-packages/pvc -name "__pycache__" -exec rm -r {} \;
+echo "Cleaning up CPython caches"
+find /usr/lib/python3/dist-packages/pvc -type d -name "__pycache__" -exec rm -fr {} + &>/dev/null || true
 
 exit 0

debian/pvc-daemon-api.preinst

@@ -1,5 +1,5 @@
 #!/bin/sh
 
 # Remove any cached CPython directories or files
-echo "Cleaning up existing CPython files"
+echo "Cleaning up CPython caches"
-find /usr/share/pvc/pvcapid -type d -name "__pycache__" -exec rm -rf {} \; &>/dev/null || true
+find /usr/share/pvc/pvcapid -type d -name "__pycache__" -exec rm -fr {} + &>/dev/null || true

debian/pvc-daemon-common.preinst

@@ -0,0 +1,5 @@
+#!/bin/sh
+
+# Remove any cached CPython directories or files
+echo "Cleaning up CPython caches"
+find /usr/share/pvc/daemon_lib -type d -name "__pycache__" -exec rm -fr {} + &>/dev/null || true

debian/pvc-daemon-health.preinst

@@ -1,6 +1,6 @@
 #!/bin/sh
 
 # Remove any cached CPython directories or files
-echo "Cleaning up existing CPython files"
+echo "Cleaning up CPython caches"
-find /usr/share/pvc/pvchealthd -type d -name "__pycache__" -exec rm -rf {} \; &>/dev/null || true
+find /usr/share/pvc/pvchealthd -type d -name "__pycache__" -exec rm -fr {} + &>/dev/null || true
-find /usr/share/pvc/plugins -type d -name "__pycache__" -exec rm -rf {} \; &>/dev/null || true
+find /usr/share/pvc/plugins -type d -name "__pycache__" -exec rm -fr {} + &>/dev/null || true

debian/pvc-daemon-node.preinst

@@ -1,5 +1,5 @@
 #!/bin/sh
 
 # Remove any cached CPython directories or files
-echo "Cleaning up existing CPython files"
+echo "Cleaning up CPython caches"
-find /usr/share/pvc/pvcnoded -type d -name "__pycache__" -exec rm -rf {} \; &>/dev/null || true
+find /usr/share/pvc/pvcnoded -type d -name "__pycache__" -exec rm -fr {} + &>/dev/null || true

debian/pvc-daemon-worker.preinst

@@ -1,5 +1,5 @@
 #!/bin/sh
 
 # Remove any cached CPython directories or files
-echo "Cleaning up existing CPython files"
+echo "Cleaning up CPython caches"
-find /usr/share/pvc/pvcworkerd -type d -name "__pycache__" -exec rm -rf {} \; &>/dev/null || true
+find /usr/share/pvc/pvcworkerd -type d -name "__pycache__" -exec rm -fr {} + &>/dev/null || true

debian/rules

@@ -13,7 +13,7 @@ override_dh_python3:
 	rm -r $(CURDIR)/client-cli/.pybuild $(CURDIR)/client-cli/pvc.egg-info
 
 override_dh_auto_clean:
-	find . -name "__pycache__" -o -name ".pybuild" -exec rm -r {} \; || true
+	find . -name "__pycache__" -o -name ".pybuild" -exec rm -fr {} + || true
 
 # If you need to rebuild the Sphinx documentation
 # Add spinxdoc to the dh --with line

health-daemon/pvchealthd/Daemon.py

@@ -33,7 +33,7 @@ import os
 import signal
 
 # Daemon version
-version = "0.9.90"
+version = "0.9.93"
 
 
 ##########################################################

node-daemon/pvcnoded/Daemon.py

@@ -49,7 +49,7 @@ import re
 import json
 
 # Daemon version
-version = "0.9.90"
+version = "0.9.93"
 
 
 ##########################################################

worker-daemon/pvcworkerd/Daemon.py

@@ -44,7 +44,7 @@ from daemon_lib.vmbuilder import (
 )
 
 # Daemon version
-version = "0.9.90"
+version = "0.9.93"
 
 
 config = cfg.get_configuration()