Bump version to 0.9.13

Ensures that the bytes_tohuman returns an integer to avoid the hacky
workaround of stripping off the B.

Adds a verification on the size of a new volume, that it is not larger
than the free space of the pool to prevent errors/excessively-large
volumes from being created.

Closes #120

README.md

@@ -20,6 +20,20 @@ To get started with PVC, please see the [About](https://parallelvirtualcluster.r
 
 ## Changelog
 
+#### v0.9.13
+
+  * Adds nicer startup messages for daemons
+  * Adds additional API field for stored_bytes to pool stats
+  * Fixes sorting issues with snapshot lists
+  * Fixes missing increment/decrement of snapshot_count on volumes
+  * Fixes bad calls in pool element API endpoints
+  * Fixes inconsistent bytes_tohuman behaviour in daemons
+  * Adds validation and maximum volume size on creation (must be smaller than the pool free space)
+
+#### v0.9.12
+
+  * Fixes a bug in the pvcnoded service unit file causing a Zookeeper startup race condition
+
 #### v0.9.11
 
   * Documentation updates

api-daemon/pvcapid/Daemon.py

@@ -26,10 +26,31 @@ import pvcapid.flaskapi as pvc_api
 # Entrypoint
 ##########################################################
 
+# Version string for startup output
+version = '0.9.11'
+
 if pvc_api.config['ssl_enabled']:
     context = (pvc_api.config['ssl_cert_file'], pvc_api.config['ssl_key_file'])
 else:
     context = None
 
-print('Starting PVC API daemon at {}:{} with SSL={}, Authentication={}'.format(pvc_api.config['listen_address'], pvc_api.config['listen_port'], pvc_api.config['ssl_enabled'], pvc_api.config['auth_enabled']))
+# Print our startup messages
+print('')
+print('|--------------------------------------------------|')
+print('|           ########  ##     ##  ######            |')
+print('|           ##     ## ##     ## ##    ##           |')
+print('|           ##     ## ##     ## ##                 |')
+print('|           ########  ##     ## ##                 |')
+print('|           ##         ##   ##  ##                 |')
+print('|           ##          ## ##   ##    ##           |')
+print('|           ##           ###     ######            |')
+print('|--------------------------------------------------|')
+print('| Parallel Virtual Cluster API daemon v{0: <11} |'.format(version))
+print('| API version: v{0: <34} |'.format(pvc_api.API_VERSION))
+print('| Listen: {0: <40} |'.format('{}:{}'.format(pvc_api.config['listen_address'], pvc_api.config['listen_port'])))
+print('| SSL: {0: <43} |'.format(str(pvc_api.config['ssl_enabled'])))
+print('| Authentication: {0: <32} |'.format(str(pvc_api.config['auth_enabled'])))
+print('|--------------------------------------------------|')
+print('')
+
 pvc_api.app.run(pvc_api.config['listen_address'], pvc_api.config['listen_port'], threaded=True, ssl_context=context)

api-daemon/pvcapid/flaskapi.py

@@ -52,16 +52,16 @@ def strtobool(stringv):
 
 # Parse the configuration file
 try:
-    pvc_config_file = os.environ['PVC_CONFIG_FILE']
+    pvcapid_config_file = os.environ['PVC_CONFIG_FILE']
 except Exception:
     print('Error: The "PVC_CONFIG_FILE" environment variable must be set before starting pvcapid.')
     exit(1)
 
-print('Starting PVC API daemon')
+print('Loading configuration from file "{}"'.format(pvcapid_config_file))
 
 # Read in the config
 try:
-    with open(pvc_config_file, 'r') as cfgfile:
+    with open(pvcapid_config_file, 'r') as cfgfile:
         o_config = yaml.load(cfgfile, Loader=yaml.BaseLoader)
 except Exception as e:
     print('ERROR: Failed to parse configuration file: {}'.format(e))
@@ -3337,12 +3337,15 @@ class API_Storage_Ceph_Pool_Root(Resource):
                     id:
                       type: integer
                       description: The Ceph pool ID
+                    stored_bytes:
+                      type: integer
+                      description: The stored data size (in bytes, post-replicas)
                     free_bytes:
                       type: integer
-                      description: The total free space (in bytes)
+                      description: The total free space (in bytes. post-replicas)
                     used_bytes:
                       type: integer
-                      description: The total used space (in bytes)
+                      description: The total used space (in bytes, pre-replicas)
                     used_percent:
                       type: number
                       description: The ratio of used space to free space
@@ -3464,7 +3467,7 @@ class API_Storage_Ceph_Pool_Element(Resource):
               type: object
               id: Message
         """
-        return api_helper, api_helper.ceph_pool_list(
+        return api_helper.ceph_pool_list(
             pool,
             is_fuzzy=False
         )
@@ -3508,7 +3511,7 @@ class API_Storage_Ceph_Pool_Element(Resource):
               type: object
               id: Message
         """
-        api_helper.ceph_pool_add(
+        return api_helper.ceph_pool_add(
             pool,
             reqargs.get('pgs', None),
             reqargs.get('replcfg', None)

api-daemon/pvcapid/ova.py

@@ -177,8 +177,8 @@ def upload_ova(pool, name, ova_size):
         pvc_common.stopZKConnection(zk_conn)
 
     # Normalize the OVA size to bytes
-    ova_size_bytes = int(pvc_ceph.format_bytes_fromhuman(ova_size)[:-1])
-    ova_size = pvc_ceph.format_bytes_fromhuman(ova_size)
+    ova_size_bytes = pvc_ceph.format_bytes_fromhuman(ova_size)
+    ova_size = '{}B'.format(ova_size_bytes)
 
     # Verify that the cluster has enough space to store the OVA volumes (2x OVA size, temporarily, 1x permanently)
     zk_conn = pvc_common.startZKConnection(config['coordinators'])
@@ -274,7 +274,7 @@ def upload_ova(pool, name, ova_size):
         vm_volume_size = disk.get('capacity')
 
         # Normalize the dev size to bytes
-        dev_size = pvc_ceph.format_bytes_fromhuman(dev_size_raw)
+        dev_size = '{}B'.format(pvc_ceph.format_bytes_fromhuman(dev_size_raw))
 
         def cleanup_img_maps():
             zk_conn = pvc_common.startZKConnection(config['coordinators'])
@@ -368,7 +368,7 @@ def upload_ova(pool, name, ova_size):
         vm_volume_size = disk.get('capacity')
 
         # The function always return XXXXB, so strip off the B and convert to an integer
-        vm_volume_size_bytes = int(pvc_ceph.format_bytes_fromhuman(vm_volume_size)[:-1])
+        vm_volume_size_bytes = pvc_ceph.format_bytes_fromhuman(vm_volume_size)
         vm_volume_size_gb = math.ceil(vm_volume_size_bytes / 1024 / 1024 / 1024)
 
         query = "INSERT INTO ova_volume (ova, pool, volume_name, volume_format, disk_id, disk_size_gb) VALUES (%s, %s, %s, %s, %s, %s);"

bump-version

@@ -17,6 +17,7 @@ $EDITOR ${changelog_file}
 changelog="$( cat ${changelog_file} | grep -v '^#' | sed 's/^*/  */' )"
 
 sed -i "s,version = '${current_version}',version = '${new_version}'," node-daemon/pvcnoded/Daemon.py
+sed -i "s,version = '${current_version}',version = '${new_version}'," api-daemon/pvcapid/Daemon.py
 
 readme_tmpdir=$( mktemp -d )
 cp README.md ${readme_tmpdir}/

daemon-common/ceph.py

@@ -122,7 +122,7 @@ def format_bytes_fromhuman(datahuman):
         dataunit = 'B'
         datasize = int(datahuman)
     databytes = datasize * byte_unit_matrix[dataunit]
-    return '{}B'.format(databytes)
+    return databytes
 
 
 # Format ops sizes to/from human-readable units
@@ -475,7 +475,17 @@ def getVolumeInformation(zk_conn, pool, volume):
 
 
 def add_volume(zk_conn, pool, name, size):
-    # 1. Create the volume
+    # 1. Verify the size of the volume
+    pool_information = getPoolInformation(zk_conn, pool)
+    size_bytes = format_bytes_fromhuman(size)
+    if size_bytes >= int(pool_information['stats']['free_bytes']):
+        return False, 'ERROR: Requested volume size is greater than the available free space in the pool'
+
+    # Add 'B' if the volume is in bytes
+    if re.match(r'^[0-9]+$', size):
+        size = '{}B'.format(size)
+
+    # 2. Create the volume
     retcode, stdout, stderr = common.run_os_command('rbd create --size {} --image-feature layering,exclusive-lock {}/{}'.format(size, pool, name))
     if retcode:
         return False, 'ERROR: Failed to create RBD volume "{}": {}'.format(name, stderr)
@@ -545,7 +555,7 @@ def resize_volume(zk_conn, pool, name, size):
             target_lv_conn = libvirt.open(dest_lv)
             target_vm_conn = target_lv_conn.lookupByName(vm_info['name'])
             if target_vm_conn:
-                target_vm_conn.blockResize(volume_id, int(format_bytes_fromhuman(size)[:-1]), libvirt.VIR_DOMAIN_BLOCK_RESIZE_BYTES)
+                target_vm_conn.blockResize(volume_id, format_bytes_fromhuman(size), libvirt.VIR_DOMAIN_BLOCK_RESIZE_BYTES)
             target_lv_conn.close()
         except Exception:
             pass
@@ -715,6 +725,16 @@ def add_snapshot(zk_conn, pool, volume, name):
         '/ceph/snapshots/{}/{}/{}/stats'.format(pool, volume, name): '{}'
     })
 
+    # 3. Update the count of snapshots on this volume
+    volume_stats_raw = zkhandler.readdata(zk_conn, '/ceph/volumes/{}/{}/stats'.format(pool, volume))
+    volume_stats = dict(json.loads(volume_stats_raw))
+    # Format the size to something nicer
+    volume_stats['snapshot_count'] = volume_stats['snapshot_count'] + 1
+    volume_stats_raw = json.dumps(volume_stats)
+    zkhandler.writedata(zk_conn, {
+        '/ceph/volumes/{}/{}/stats'.format(pool, volume): volume_stats_raw
+    })
+
     return True, 'Created RBD snapshot "{}" of volume "{}" in pool "{}".'.format(name, volume, pool)
 
 
@@ -751,6 +771,16 @@ def remove_snapshot(zk_conn, pool, volume, name):
     # 2. Delete snapshot from Zookeeper
     zkhandler.deletekey(zk_conn, '/ceph/snapshots/{}/{}/{}'.format(pool, volume, name))
 
+    # 3. Update the count of snapshots on this volume
+    volume_stats_raw = zkhandler.readdata(zk_conn, '/ceph/volumes/{}/{}/stats'.format(pool, volume))
+    volume_stats = dict(json.loads(volume_stats_raw))
+    # Format the size to something nicer
+    volume_stats['snapshot_count'] = volume_stats['snapshot_count'] - 1
+    volume_stats_raw = json.dumps(volume_stats)
+    zkhandler.writedata(zk_conn, {
+        '/ceph/volumes/{}/{}/stats'.format(pool, volume): volume_stats_raw
+    })
+
     return True, 'Removed RBD snapshot "{}" of volume "{}" in pool "{}".'.format(name, volume, pool)
 
 
@@ -783,4 +813,4 @@ def get_list_snapshot(zk_conn, pool, volume, limit, is_fuzzy=True):
         else:
             snapshot_list.append({'pool': pool_name, 'volume': volume_name, 'snapshot': snapshot_name})
 
-    return True, sorted(snapshot_list, key=lambda x: int(x['id']))
+    return True, sorted(snapshot_list, key=lambda x: str(x['snapshot']))

debian/changelog

@@ -1,3 +1,21 @@
+pvc (0.9.13-0) unstable; urgency=high
+
+  * Adds nicer startup messages for daemons
+  * Adds additional API field for stored_bytes to pool stats
+  * Fixes sorting issues with snapshot lists
+  * Fixes missing increment/decrement of snapshot_count on volumes
+  * Fixes bad calls in pool element API endpoints
+  * Fixes inconsistent bytes_tohuman behaviour in daemons
+  * Adds validation and maximum volume size on creation (must be smaller than the pool free space)
+
+ -- Joshua M. Boniface <joshua@boniface.me>  Wed, 17 Feb 2021 11:33:28 -0500
+
+pvc (0.9.12-0) unstable; urgency=high
+
+  * Fixes a bug in the pvcnoded service unit file causing a Zookeeper startup race condition
+
+ -- Joshua M. Boniface <joshua@boniface.me>  Thu, 28 Jan 2021 16:29:58 -0500
+
 pvc (0.9.11-0) unstable; urgency=high
 
   * Documentation updates

docs/index.md

@@ -18,6 +18,20 @@ To get started with PVC, please see the [About](https://parallelvirtualcluster.r
 
 ## Changelog
 
+#### v0.9.13
+
+  * Adds nicer startup messages for daemons
+  * Adds additional API field for stored_bytes to pool stats
+  * Fixes sorting issues with snapshot lists
+  * Fixes missing increment/decrement of snapshot_count on volumes
+  * Fixes bad calls in pool element API endpoints
+  * Fixes inconsistent bytes_tohuman behaviour in daemons
+  * Adds validation and maximum volume size on creation (must be smaller than the pool free space)
+
+#### v0.9.12
+
+  * Fixes a bug in the pvcnoded service unit file causing a Zookeeper startup race condition
+
 #### v0.9.11
 
   * Documentation updates

docs/manuals/swagger.json

@@ -621,7 +621,7 @@
                 "stats": {
                     "properties": {
                         "free_bytes": {
-                            "description": "The total free space (in bytes)",
+                            "description": "The total free space (in bytes. post-replicas)",
                             "type": "integer"
                         },
                         "id": {
@@ -660,8 +660,12 @@
                             "description": "The total read operations on the pool (pool-lifetime)",
                             "type": "integer"
                         },
+                        "stored_bytes": {
+                            "description": "The stored data size (in bytes, post-replicas)",
+                            "type": "integer"
+                        },
                         "used_bytes": {
-                            "description": "The total used space (in bytes)",
+                            "description": "The total used space (in bytes, pre-replicas)",
                             "type": "integer"
                         },
                         "used_percent": {

node-daemon/pvcnoded.service

@@ -2,8 +2,7 @@
 
 [Unit]
 Description = Parallel Virtual Cluster node daemon
-After = network-online.target zookeeper.service libvirtd.service
-Wants = zookeeper.service libvirtd.service
+After = network-online.target
 PartOf = pvc.target
 
 [Service]

node-daemon/pvcnoded/Daemon.py

@@ -54,7 +54,7 @@ import pvcnoded.CephInstance as CephInstance
 import pvcnoded.MetadataAPIInstance as MetadataAPIInstance
 
 # Version string for startup output
-version = '0.9.11'
+version = '0.9.13'
 
 ###############################################################################
 # PVCD - node daemon startup program
@@ -134,7 +134,7 @@ def readConfig(pvcnoded_config_file, myhostname):
 
     with open(pvcnoded_config_file, 'r') as cfgfile:
         try:
-            o_config = yaml.load(cfgfile)
+            o_config = yaml.load(cfgfile, Loader=yaml.SafeLoader)
         except Exception as e:
             print('ERROR: Failed to parse configuration file: {}'.format(e))
             exit(1)
@@ -331,16 +331,29 @@ if not os.path.exists(config['log_directory']):
 logger = log.Logger(config)
 
 # Print our startup messages
-logger.out('Parallel Virtual Cluster node daemon v{}'.format(version))
-logger.out('FQDN: {}'.format(myfqdn))
-logger.out('Host: {}'.format(myhostname))
-logger.out('ID: {}'.format(mynodeid))
-logger.out('IPMI hostname: {}'.format(config['ipmi_hostname']))
-logger.out('Machine details:')
-logger.out('  CPUs: {}'.format(staticdata[0]))
-logger.out('  Arch: {}'.format(staticdata[3]))
-logger.out('  OS: {}'.format(staticdata[2]))
-logger.out('  Kernel: {}'.format(staticdata[1]))
+logger.out('')
+logger.out('|--------------------------------------------------|')
+logger.out('|           ########  ##     ##  ######            |')
+logger.out('|           ##     ## ##     ## ##    ##           |')
+logger.out('|           ##     ## ##     ## ##                 |')
+logger.out('|           ########  ##     ## ##                 |')
+logger.out('|           ##         ##   ##  ##                 |')
+logger.out('|           ##          ## ##   ##    ##           |')
+logger.out('|           ##           ###     ######            |')
+logger.out('|--------------------------------------------------|')
+logger.out('| Parallel Virtual Cluster node daemon v{0: <10} |'.format(version))
+logger.out('| FQDN: {0: <42} |'.format(myfqdn))
+logger.out('| Host: {0: <42} |'.format(myhostname))
+logger.out('| ID: {0: <44} |'.format(mynodeid))
+logger.out('| IPMI hostname: {0: <33} |'.format(config['ipmi_hostname']))
+logger.out('| Machine details:                                 |')
+logger.out('|   CPUs: {0: <40} |'.format(staticdata[0]))
+logger.out('|   Arch: {0: <40} |'.format(staticdata[3]))
+logger.out('|   OS: {0: <42} |'.format(staticdata[2]))
+logger.out('|   Kernel: {0: <38} |'.format(staticdata[1]))
+logger.out('|--------------------------------------------------|')
+logger.out('')
+
 logger.out('Starting pvcnoded on host {}'.format(myfqdn), state='s')
 
 # Define some colours for future messages if applicable
@@ -1142,6 +1155,7 @@ def collect_ceph_stats(queue):
                 # Assemble a useful data structure
                 pool_df = {
                     'id': pool['id'],
+                    'stored_bytes': pool['stats']['stored'],
                     'free_bytes': pool['stats']['max_avail'],
                     'used_bytes': pool['stats']['bytes_used'],
                     'used_percent': pool['stats']['percent_used'],