Increase ipmi plugin timeout

Makes some places clearer, cleans up cruft, and adds references to the
main documentation as required.

.version

@@ -1 +1 @@
-0.9.74
+0.9.78

CHANGELOG.md

@@ -1,5 +1,24 @@
 ## PVC Changelog
 
+###### [v0.9.78](https://github.com/parallelvirtualcluster/pvc/releases/tag/v0.9.78)
+
+  * [API, Client CLI] Fixes several bugs around image uploads; adds a new query parameter for non-raw images
+  * [API] Ensures RBD images are created with a raw bytes value to avoid rounding errors
+
+###### [v0.9.77](https://github.com/parallelvirtualcluster/pvc/releases/tag/v0.9.77)
+
+  * [Client CLI] Fixes a bug from a bad library import
+
+###### [v0.9.76](https://github.com/parallelvirtualcluster/pvc/releases/tag/v0.9.76)
+
+  * [API, Client CLI] Corrects some missing node states for fencing in status output
+
+###### [v0.9.75](https://github.com/parallelvirtualcluster/pvc/releases/tag/v0.9.75)
+
+  * [Node Daemon] Adds a startup message about IPMI when succeeding
+  * [Node Daemon] Fixes a bug in fencing allowing non-failing VMs to migrate
+  * [Node Daemon] Adds rounding to load average in load plugin for consistency
+
 ###### [v0.9.74](https://github.com/parallelvirtualcluster/pvc/releases/tag/v0.9.74)
 
   * [Docs] Removes docs from the main repo

api-daemon/pvcapid/Daemon.py

@@ -27,7 +27,7 @@ from ssl import SSLContext, TLSVersion
 from distutils.util import strtobool as dustrtobool
 
 # Daemon version
-version = "0.9.74"
+version = "0.9.78"
 
 # API version
 API_VERSION = 1.0

api-daemon/pvcapid/flaskapi.py

@@ -4843,7 +4843,7 @@ class API_Storage_Ceph_Volume_Root(Resource):
             {
                 "name": "size",
                 "required": True,
-                "helptext": "A volume size in bytes (or with k/M/G/T suffix) must be specified.",
+                "helptext": "A volume size in bytes (B implied or with SI suffix k/M/G/T) must be specified.",
             },
         ]
     )
@@ -4869,7 +4869,7 @@ class API_Storage_Ceph_Volume_Root(Resource):
             name: size
             type: string
             required: true
-            description: The volume size in bytes (or with a metric suffix, i.e. k/M/G/T)
+            description: The volume size, in bytes (B implied) or with a single-character SI suffix (k/M/G/T)
         responses:
           200:
             description: OK
@@ -5088,7 +5088,12 @@ class API_Storage_Ceph_Volume_Element_Upload(Resource):
                 "required": True,
                 "location": ["args"],
                 "helptext": "A source image format must be specified.",
-            }
+            },
+            {
+                "name": "file_size",
+                "required": False,
+                "location": ["args"],
+            },
         ]
     )
     @Authenticator
@@ -5113,6 +5118,11 @@ class API_Storage_Ceph_Volume_Element_Upload(Resource):
               - qed
               - vdi
               - vpc
+          - in: query
+            name: file_size
+            type: integer
+            required: false
+            description: The size of the image file, in bytes, if {image_format} is not "raw"
         responses:
           200:
             description: OK
@@ -5131,7 +5141,10 @@ class API_Storage_Ceph_Volume_Element_Upload(Resource):
               id: Message
         """
         return api_helper.ceph_volume_upload(
-            pool, volume, reqargs.get("image_format", None)
+            pool,
+            volume,
+            reqargs.get("image_format", None),
+            reqargs.get("file_size", None),
         )
 
 

api-daemon/pvcapid/helper.py

@@ -1584,7 +1584,7 @@ def ceph_volume_remove(zkhandler, pool, name):
 
 
 @ZKConnection(config)
-def ceph_volume_upload(zkhandler, pool, volume, img_type):
+def ceph_volume_upload(zkhandler, pool, volume, img_type, file_size=None):
     """
     Upload a raw file via HTTP post to a PVC Ceph volume
     """
@@ -1605,7 +1605,17 @@ def ceph_volume_upload(zkhandler, pool, volume, img_type):
         }
         retcode = 400
         return output, retcode
-    dev_size = retdata[0]["stats"]["size"]
+
+    try:
+        dev_size = retdata[0]["stats"]["size"]
+    except Exception:
+        output = {
+            "message": "Target volume '{}' does not exist in pool '{}'.".format(
+                volume, pool
+            )
+        }
+        retcode = 400
+        return output, retcode
 
     def cleanup_maps_and_volumes():
         # Unmap the target blockdev
@@ -1619,8 +1629,14 @@ def ceph_volume_upload(zkhandler, pool, volume, img_type):
             zkhandler, pool, "{}_tmp".format(volume)
         )
 
-    # Create a temporary block device to store non-raw images
     if img_type == "raw":
+        if file_size != dev_size:
+            output = {
+                "message": f"Image file size {file_size} does not match volume size {dev_size}"
+            }
+            retcode = 400
+            return output, retcode
+
         # Map the target blockdev
         retflag, retdata = pvc_ceph.map_volume(zkhandler, pool, volume)
         if not retflag:
@@ -1659,11 +1675,15 @@ def ceph_volume_upload(zkhandler, pool, volume, img_type):
         cleanup_maps_and_volumes()
         return output, retcode
 
-    # Write the image directly to the blockdev
     else:
+        if file_size is None:
+            output = {"message": "A file size must be specified"}
+            retcode = 400
+            return output, retcode
+
         # Create a temporary blockdev
         retflag, retdata = pvc_ceph.add_volume(
-            zkhandler, pool, "{}_tmp".format(volume), dev_size
+            zkhandler, pool, "{}_tmp".format(volume), file_size
         )
         if not retflag:
             output = {"message": retdata.replace('"', "'")}

client-cli/pvc/cli/cli.py

@@ -1188,10 +1188,8 @@ def cli_vm_modify(
 
     # Grab the current config
     current_vm_cfg_raw = vm_information.get("xml")
-    xml_data = etree.fromstring(current_vm_cfg_raw)
-    current_vm_cfgfile = (
-        etree.tostring(xml_data, pretty_print=True).decode("utf8").strip()
-    )
+    xml_data = fromstring(current_vm_cfg_raw)
+    current_vm_cfgfile = tostring(xml_data, pretty_print=True).decode("utf8").strip()
 
     if editor is True:
         new_vm_cfgfile = click.edit(
@@ -3459,14 +3457,14 @@ def cli_storage_pool():
     show_default=True,
     required=False,
     help="""
-    The replication configuration, specifying both a "copies" and "mincopies" value, separated by a comma, e.g. "copies=3,mincopies=2". The "copies" value specifies the total number of replicas and should not exceed the total number of nodes; the "mincopies" value specifies the minimum number of available copies to allow writes. For additional details please see the Cluster Architecture documentation.
+    The replication configuration, specifying both a "copies" and "mincopies" value, separated by a comma, e.g. "copies=3,mincopies=2". The "copies" value specifies the total number of replicas and the "mincopies" value specifies the minimum number of active replicas to allow I/O. For additional details please see the documentation.
     """,
 )
 def cli_storage_pool_add(name, pgs, tier, replcfg):
     """
     Add a new Ceph RBD pool with name NAME and PGS placement groups.
 
-    The placement group count must be a non-zero power of 2.
+    The placement group count must be a non-zero power of 2. Generally you should choose a PGS number such that there will be 50-150 PGs on each OSD in a single node (before replicas); 64, 128, or 256 are good values for small clusters (1-5 OSDs per node); higher values are recommended for higher node or OSD counts. For additional details please see the documentation.
     """
 
     retcode, retmsg = pvc.lib.storage.ceph_pool_add(
@@ -3505,9 +3503,9 @@ def cli_storage_pool_set_pgs(name, pgs):
     """
     Set the placement groups (PGs) count for the pool NAME to PGS.
 
-    The placement group count must be a non-zero power of 2.
+    The placement group count must be a non-zero power of 2. Generally you should choose a PGS number such that there will be 50-150 PGs on each OSD in a single node (before replicas); 64, 128, or 256 are good values for small clusters (1-5 OSDs per node); higher values are recommended for higher node or OSD counts. For additional details please see the documentation.
 
-    Placement group counts may be increased or decreased as required though frequent alteration is not recommended.
+    Placement group counts may be increased or decreased as required though frequent alteration is not recommended. Placement group alterations are intensive operations on the storage cluster.
     """
 
     retcode, retmsg = pvc.lib.storage.ceph_pool_set_pgs(CLI_CONFIG, name, pgs)
@@ -3600,7 +3598,7 @@ def cli_storage_volume_upload(pool, name, image_format, image_file):
     If the image format is "raw", the image is uploaded directly to the target volume without modification. Otherwise, it will be converted into raw format by "qemu-img convert" on the remote side before writing using a temporary volume. The image format must be a valid format recognized by "qemu-img", such as "vmdk" or "qcow2".
     """
 
-    if not os.path.exists(image_file):
+    if not path.exists(image_file):
         echo(CLI_CONFIG, "ERROR: File '{}' does not exist!".format(image_file))
         exit(1)
 
@@ -4912,13 +4910,13 @@ def cli_provisioner_ova_upload(name, filename, pool):
     Storage templates, provisioning scripts, and arguments for OVA-type profiles will be ignored and should not be set.
     """
 
-    if not os.path.exists(filename):
+    if not path.exists(filename):
         echo(CLI_CONFIG, "ERROR: File '{}' does not exist!".format(filename))
         exit(1)
 
     params = dict()
     params["pool"] = pool
-    params["ova_size"] = os.path.getsize(filename)
+    params["ova_size"] = path.getsize(filename)
 
     retcode, retdata = pvc.lib.provisioner.ova_upload(
         CLI_CONFIG, name, filename, params

client-cli/pvc/cli/formatters.py

@@ -135,7 +135,7 @@ def cli_cluster_status_format_pretty(CLI_CONFIG, data):
             state_colour = ansii["green"]
         elif state in ["run,flush", "run,unflush", "run,flushed"]:
             state_colour = ansii["blue"]
-        elif "dead" in state or "stop" in state:
+        elif "dead" in state or "fenced" in state or "stop" in state:
             state_colour = ansii["red"]
         else:
             state_colour = ansii["yellow"]

client-cli/pvc/lib/storage.py

@@ -21,6 +21,7 @@
 
 import math
 
+from os import path
 from json import loads
 from requests_toolbelt.multipart.encoder import (
     MultipartEncoder,
@@ -1209,6 +1210,11 @@ def ceph_volume_upload(config, pool, volume, image_format, image_file):
     """
     import click
 
+    if image_format != "raw":
+        file_size = path.getsize(image_file)
+    else:
+        file_size = None
+
     bar = UploadProgressBar(
         image_file, end_message="Parsing file on remote side...", end_nl=False
     )
@@ -1220,7 +1226,7 @@ def ceph_volume_upload(config, pool, volume, image_format, image_file):
     upload_monitor = MultipartEncoderMonitor(upload_data, bar.update)
 
     headers = {"Content-Type": upload_monitor.content_type}
-    params = {"image_format": image_format}
+    params = {"image_format": image_format, "file_size": file_size}
 
     response = call_api(
         config,

client-cli/setup.py

@@ -2,7 +2,7 @@ from setuptools import setup
 
 setup(
     name="pvc",
-    version="0.9.74",
+    version="0.9.78",
     packages=["pvc.cli", "pvc.lib"],
     install_requires=[
         "Click",

daemon-common/ceph.py

@@ -763,9 +763,7 @@ def add_volume(zkhandler, pool, name, size):
 
     # 2. Create the volume
     retcode, stdout, stderr = common.run_os_command(
-        "rbd create --size {} {}/{}".format(
-            format_bytes_tohuman(size_bytes), pool, name
-        )
+        "rbd create --size {}B {}/{}".format(size_bytes, pool, name)
     )
     if retcode:
         return False, 'ERROR: Failed to create RBD volume "{}": {}'.format(name, stderr)

daemon-common/cluster.py

@@ -256,8 +256,13 @@ def getClusterInformation(zkhandler):
         "stop,unflush",
         "dead,ready",
         "dead,flush",
+        "dead,fence-flush",
         "dead,flushed",
         "dead,unflush",
+        "fenced,ready",
+        "fenced,flush",
+        "fenced,flushed",
+        "fenced,unflush",
     ]
     vm_state_combinations = [
         "start",

debian/changelog

@@ -1,3 +1,30 @@
+pvc (0.9.78-0) unstable; urgency=high
+
+  * [API, Client CLI] Fixes several bugs around image uploads; adds a new query parameter for non-raw images
+  * [API] Ensures RBD images are created with a raw bytes value to avoid rounding errors
+
+ -- Joshua M. Boniface <joshua@boniface.me>  Sat, 30 Sep 2023 12:57:55 -0400
+
+pvc (0.9.77-0) unstable; urgency=high
+
+  * [Client CLI] Fixes a bug from a bad library import
+
+ -- Joshua M. Boniface <joshua@boniface.me>  Tue, 19 Sep 2023 11:05:55 -0400
+
+pvc (0.9.76-0) unstable; urgency=high
+
+  * [API, Client CLI] Corrects some missing node states for fencing in status output
+
+ -- Joshua M. Boniface <joshua@boniface.me>  Mon, 18 Sep 2023 10:15:52 -0400
+
+pvc (0.9.75-0) unstable; urgency=high
+
+  * [Node Daemon] Adds a startup message about IPMI when succeeding
+  * [Node Daemon] Fixes a bug in fencing allowing non-failing VMs to migrate
+  * [Node Daemon] Adds rounding to load average in load plugin for consistency
+
+ -- Joshua M. Boniface <joshua@boniface.me>  Sat, 16 Sep 2023 23:06:38 -0400
+
 pvc (0.9.74-0) unstable; urgency=high
 
   * [Docs] Removes docs from the main repo

gen-api-doc

@@ -14,7 +14,7 @@ sys.path.append('api-daemon')
 
 import pvcapid.flaskapi as pvc_api
 
-swagger_file = "docs/manuals/swagger.json"
+swagger_file = "swagger.json"
 swagger_data = swagger(pvc_api.app)
 swagger_data['info']['version'] = "1.0"
 swagger_data['info']['title'] = "PVC Client and Provisioner API"
@@ -22,3 +22,5 @@ swagger_data['host'] = "pvc.local:7370"
 
 with open(swagger_file, 'w') as fd:
     fd.write(json.dumps(swagger_data, sort_keys=True, indent=4))
+
+print(f"Swagger file output to {swagger_file}; add it to the PVC documentation repo.")

node-daemon/plugins/ipmi

@@ -76,7 +76,7 @@ class MonitoringPluginScript(MonitoringPlugin):
         ipmi_password = self.config["ipmi_password"]
         retcode, _, _ = run_os_command(
             f"/usr/bin/ipmitool -I lanplus -H {ipmi_hostname} -U {ipmi_username} -P {ipmi_password} chassis power status",
-            timeout=2
+            timeout=5
         )
 
         if retcode > 0:

node-daemon/plugins/load

@@ -72,7 +72,7 @@ class MonitoringPluginScript(MonitoringPlugin):
         from psutil import cpu_count
 
         # Get the current 1-minute system load average
-        load_average = getloadavg()[0]
+        load_average = float(round(getloadavg()[0], 2))
 
         # Get the number of CPU cores
         cpu_cores = cpu_count()

node-daemon/pvcnoded/Daemon.py

@@ -49,7 +49,7 @@ import re
 import json
 
 # Daemon version
-version = "0.9.74"
+version = "0.9.78"
 
 
 ##########################################################
@@ -324,9 +324,14 @@ def entrypoint():
         config["ipmi_hostname"], config["ipmi_username"], config["ipmi_password"]
     ):
         logger.out(
-            "Our IPMI is not reachable; fencing of this node will likely fail",
+            "Our IPMI interface is not reachable; fencing of this node will fail until corrected",
             state="w",
         )
+    else:
+        logger.out(
+            "Our IPMI interface is reachable; fencing of this node is possible",
+            state="o",
+        )
 
     # Validate libvirt
     if not pvcnoded.util.libvirt.validate_libvirtd(logger, config):

node-daemon/pvcnoded/util/fencing.py

@@ -153,7 +153,13 @@ def migrateFromFencedNode(zkhandler, node_name, config, logger):
 
     # Loop through the VMs
     for dom_uuid in dead_node_running_domains:
-        fence_migrate_vm(dom_uuid)
+        try:
+            fence_migrate_vm(dom_uuid)
+        except Exception as e:
+            logger.out(
+                f"Failed to migrate VM {dom_uuid}, continuing: {e}",
+                state="w",
+            )
 
     # Set node in flushed state for easy remigrating when it comes back
     zkhandler.write([(("node.state.domain", node_name), "flushed")])