community/front-end/ofe/website/ghpcfe/cluster_manager/cloud_info.py

# Copyright 2022 Google LLC
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#      http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

"""Cloud interrogation routines"""

import json
import logging
import time
from collections import defaultdict
from functools import lru_cache

import archspec.cpu
import google.cloud.exceptions
import googleapiclient.discovery
from google.cloud import storage as gcs
from google.cloud.billing_v1.services import cloud_catalog
from google.oauth2 import service_account

logger = logging.getLogger(__name__)

gcp_machine_table = defaultdict(
    lambda: defaultdict(lambda: "x86_64"),
    {
        # General Purpose
        "e2": defaultdict(lambda: "x86_64"),
        "n2": defaultdict(lambda: "cascadelake"),
        "n2d": defaultdict(lambda: "zen2"),
        "n1": defaultdict(lambda: "x86_64"),
        "c3": defaultdict(lambda: "sapphirerapids"),
        "c3d": defaultdict(lambda: "zen2"),
        "c4": defaultdict(lambda: "emeraldrapids"),
        # Compute Optimized
        "c2": defaultdict(lambda: "cascadelake"),
        "c2d": defaultdict(
            lambda: "zen2"  # TODO: Should be zen3, but CentOS7 doesn't have
        ),  # a new enough kernel to recognize as such.
        "t2d": defaultdict(lambda: "zen2"),  # TODO: Should also be zen3
        "h3": defaultdict(lambda: "sapphirerapids"),
        # Memory Optimized
        "m2": defaultdict(lambda: "icelake"),
        "m2": defaultdict(lambda: "cascadelake"),
        "m1": defaultdict(
            lambda: "broadwell",
            {"megamem": "skylake_avx512", "ultramem": "broadwell"},
        ),
        # Accelerated
        "a2": defaultdict(lambda: "cascadelake"),
    },
)


def _get_arch_for_node_type_gcp(instance):
    try:
        family, group, _ = instance.split("-", maxsplit=2)
        return gcp_machine_table[family][group]
    except ValueError:
        logger.error(f"Invalid instance format: {instance}")
        return None
    except KeyError:
        logger.error(f"Keys not found in gcp_machine_table: {instance}")
        return None


def _get_gcp_client(credentials, service="compute", api_version="v1"):
    cred_info = json.loads(credentials)
    creds = service_account.Credentials.from_service_account_info(cred_info)
    return (
        cred_info["project_id"],
        googleapiclient.discovery.build(
            service, api_version, credentials=creds, cache_discovery=False
        ),
    )


def _get_vm_reservations(credentials, zone, ttl_hash=None):
    try:
        # logger.info(f"Fetching VM reservations for credentials: {credentials}, zone: {zone}")
        project, client = _get_gcp_client(credentials)

        req = client.reservations().list(project=project, zone=zone)
        resp = req.execute()

        if "items" not in resp:
            # logger.info("No reservations found")
            return {}

        data = {
            reservation["name"]: {
                "name": reservation["name"],
                "specificReservationRequired": reservation.get("specificReservationRequired", False),
                "status": reservation["status"],
                "instanceProperties": {
                    "machineType": reservation
                        .get("specificReservation", {})
                        .get("instanceProperties", {})
                        .get("machineType", ""),
                    "minCpuPlatform": reservation
                        .get("specificReservation", {})
                        .get("instanceProperties", {})
                        .get("minCpuPlatform", ""),
                    "availableCount": int(
                        reservation
                            .get("specificReservation", {})
                            .get("count", 0)
                    )
                },
                "shareSettings": reservation.get("shareSettings", {}),
            }
            for reservation in resp["items"]
        }

        # logger.info(f"Reservations data: {data}")
        return data
    except Exception as e:
        logger.error(f"Error fetching VM reservations: {e}")
        return {}

def get_vm_reservations(cloud_provider, credentials, unused_region, zone):
    if cloud_provider == "GCP":
        return _get_vm_reservations(credentials, zone, ttl_hash=_get_ttl_hash())
    else:
        raise Exception(f'Unsupported Cloud Provider "{cloud_provider}"')


@lru_cache
def _get_gcp_disk_types(
    credentials, zone, ttl_hash=None
):  # pylint: disable=unused-argument
    (project, client) = _get_gcp_client(credentials)

    req = client.diskTypes().list(project=project, zone=zone)
    resp = req.execute()
    return [
        {
            "description": x["description"],
            "name": x["name"],
            "minSizeGB": int(x["validDiskSize"].split("-")[0][:-2]),
            "maxSizeGB": int(x["validDiskSize"].split("-")[1][:-2]),
        }
        for x in resp.get("items", [])
    ]

def get_disk_types(cloud_provider, credentials, unused_region, zone):
    if cloud_provider == "GCP":
        return _get_gcp_disk_types(
            credentials, zone, ttl_hash=_get_ttl_hash()
        )
    else:
        raise Exception(f'Unsupport Cloud Provider "{cloud_provider}"')


@lru_cache
def _get_gcp_machine_types(
    credentials, zone, ttl_hash=None
):  # pylint: disable=unused-argument
    (project, client) = _get_gcp_client(credentials)

    # Fetch disk types dynamically
    disk_types = _get_gcp_disk_types(credentials, zone, ttl_hash=ttl_hash)
    disk_type_names = [disk_type["name"] for disk_type in disk_types]

    req = client.machineTypes().list(
        project=project, zone=zone, filter="isSharedCpu=False"
    )

    resp = req.execute()
    if "items" not in resp:
        return []

    invalid_disk_types = {
        "c4-": [
            "local-ssd", "pd-standard", "pd-balanced", "pd-ssd",
            "pd-extreme", "hyperdisk-ml", "hyperdisk-throughput"
        ],
        "c3-": [
            "pd-extreme", "pd-standard"
        ],
        "c3d-": [
            "pd-standard", "pd-extreme", "hyperdisk-extreme"
        ],
        "n4-": [
            "local-ssd", "pd-standard", "pd-balanced", "pd-ssd",
            "pd-extreme", "hyperdisk-extreme", "hyperdisk-ml",
            "hyperdisk-throughput"
        ],
        "n2-": [
            "hyperdisk-balanced", "hyperdisk-ml"
        ],
        "n2d-": [
            "pd-extreme", "hyperdisk-ml", "hyperdisk-balanced",
            "hyperdisk-extreme"
        ],
        "n1-": [
            "pd-extreme", "hyperdisk-extreme", "hyperdisk-ml",
            "hyperdisk-throughput", "hyperdisk-balanced"
        ],
        "t2d-": [
            "pd-extreme", "local-ssd", "hyperdisk-balanced",
            "hyperdisk-ml", "hyperdisk-extreme"
        ],
        "t2a-": [
            "local-ssd", "pd-extreme", "hyperdisk-balanced",
            "hyperdisk-ml", "hyperdisk-extreme",
            "hyperdisk-throughput"
        ],
        "e2-": [
            "local-ssd", "pd-extreme", "hyperdisk-balanced",
            "hyperdisk-ml", "hyperdisk-extreme",
            "hyperdisk-throughput"
        ],
        "z3-": [
            "pd-extreme", "pd-standard", "hyperdisk-balanced",
            "hyperdisk-ml"
        ],
        "h3-": [
            "local-ssd", "pd-standard", "pd-ssd", "pd-extreme",
            "hyperdisk-ml", "hyperdisk-extreme"
        ],
        "c2-": [
            "pd-extreme", "hyperdisk-balanced", "hyperdisk-extreme",
            "hyperdisk-ml", "hyperdisk-throughput"
        ],
        "c2d-": [
            "pd-extreme", "hyperdisk-balanced", "hyperdisk-extreme",
            "hyperdisk-ml", "hyperdisk-throughput"
        ],
        "x4-": [
            "local-ssd", "pd-ssd", "pd-standard", "pd-balanced",
            "pd-extreme", "hyperdisk-ml", "hyperdisk-throughput"
        ],
        "m3-": [
            "hyperdisk-throughput", "hyperdisk-ml", "pd-standard"
        ],
        "m2-": [
            "local-ssd", "hyperdisk-ml", "hyperdisk-throughput"
        ],
        "m1-": [
            "hyperdisk-ml", "hyperdisk-throughput"
        ],
        "n1-": [
            "pd-extreme", "hyperdisk-balanced", "hyperdisk-ml",
            "hyperdisk-extreme", "hyperdisk-throughput"
        ],
        "a3-": [
            "pd-extreme", "pd-standard",
            "hyperdisk-balanced"
        ],
        "a2-": [
            "pd-extreme", "hyperdisk-throughput",
            "hyperdisk-balanced", "hyperdisk-extreme"
        ],
        "g2-": [
            "pd-extreme", "pd-standard", "hyperdisk-balanced",
            "hyperdisk-extreme"
        ]
    }

    def get_invalid_disk_types(machine_type_name):
        family = machine_type_name.split("-")[0] + "-"
        return invalid_disk_types.get(family, [])

    data = {
        mt["name"]: {
            "name": mt["name"],
            "family": mt["name"].split("-")[0],
            "memory": mt["memoryMb"],
            "vCPU": mt["guestCpus"],
            "arch": _get_arch_for_node_type_gcp(mt["name"]),
            "accelerators": {
                acc["guestAcceleratorType"]: {
                    "min_count": acc["guestAcceleratorCount"],
                    "max_count": acc["guestAcceleratorCount"],
                }
                for acc in mt.get("accelerators", [])
            },
            "invalid_disk_types": get_invalid_disk_types(mt["name"])
        }
        for mt in resp["items"]
    }

    # Grab the Accelerators
    accels = (
        client.acceleratorTypes()
        .list(project=project, zone=zone)
        .execute()
    )
    # Set N1-associated Accelerators
    n1_accels = {
        acc["name"]: {
            "description": acc["description"],
            "min_count": 0,
            "max_count": acc["maximumCardsPerInstance"],
        }
        for acc in accels.get("items", [])
        if "nvidia-tesla-a100" not in acc["name"]
    }
    for mach in data.keys():
        if data[mach]["family"] == "n1":
            data[mach]["accelerators"] = n1_accels
        # Fix up description for A100 (or others)
        elif data[mach]["accelerators"]:
            for acc_name in data[mach]["accelerators"].keys():
                items = [
                    x
                    for x in accels.get("items", [])
                    if x["name"] == acc_name
                ]
                if items:
                    data[mach]["accelerators"][acc_name]["description"] = (
                        items[0]["description"]
                    )

    # logger.info(data)

    return data


def _get_ttl_hash(seconds=3600 * 24):
    """Return the same value within `seconds` time period.

    Default to 1 day of caching
    """
    return round(time.time() / seconds)


def get_machine_types(cloud_provider, credentials, unused_region, zone):
    if cloud_provider == "GCP":
        return _get_gcp_machine_types(
            credentials, zone, ttl_hash=_get_ttl_hash()
        )
    else:
        raise Exception(f'Unsupport Cloud Provider "{cloud_provider}"')


def _get_arch_ancestry(arch):
    ancestry = {arch.name}
    for p in arch.parents:
        ancestry.update(_get_arch_ancestry(p))
    return ancestry


def get_common_arch(archs):
    archs = [archspec.cpu.TARGETS[a] for a in archs]
    common_arch_set = set.intersection(*[_get_arch_ancestry(a) for a in archs])
    if not common_arch_set:
        return None
    return max([archspec.cpu.TARGETS[a] for a in common_arch_set]).name


def get_arch_ancestry(arch_name):
    arch = archspec.cpu.TARGETS[arch_name]
    res = [arch_name]
    if arch.family != arch:
        for x in arch.parents:
            res.extend(get_arch_ancestry(x.name))
    return res


def get_arch_family(arch):
    return archspec.cpu.TARGETS[arch].family.name


def sort_architectures(arch_names):
    archs = [archspec.cpu.TARGETS[a] for a in arch_names]
    return [x.name for x in sorted(archs)]


@lru_cache
def _get_gcp_region_zone_info(
    credentials, ttl_hash=None
):  # pylint: disable=unused-argument
    (project, client) = _get_gcp_client(credentials)

    req = client.zones().list(project=project)
    results = defaultdict(list)
    while req is not None:
        resp = req.execute()
        for zone in resp["items"]:
            region = "-".join(zone["name"].split("-")[:-1])
            results[region].append(zone["name"])
        req = client.zones().list_next(
            previous_request=req, previous_response=resp
        )
    return results


def get_region_zone_info(cloud_provider, credentials):
    if cloud_provider == "GCP":
        return _get_gcp_region_zone_info(credentials, ttl_hash=_get_ttl_hash())
    else:
        raise Exception("Unsupported Cloud Provider")


def _get_gcp_subnets(credentials):
    (project, client) = _get_gcp_client(credentials)

    req = client.subnetworks().listUsable(project=project)
    results = req.execute()
    entries = results["items"]
    subnets = []
    for entry in entries:
        # subnet in the form of https://www.googleapis.com/compute/v1/projects/<project>/regions/<region>/subnetworks/<name>
        tokens = entry["subnetwork"].split("/")
        region = tokens[8]
        subnet = tokens[10]
        # vpc in the form of https://www.googleapis.com/compute/v1/projects/<project>/global/networks/<name>
        tokens = entry["network"].split("/")
        vpc = tokens[9]
        # cidr in standard form xxx.xxx.xxx.xxx/yy
        cidr = entry["ipCidrRange"]
        subnets.append([vpc, region, subnet, cidr])
    return subnets


def get_subnets(cloud_provider, credentials):
    if cloud_provider == "GCP":
        return _get_gcp_subnets(credentials)
    else:
        raise Exception("Unsupported Cloud Provider")


_gcp_services_list = None
_gcp_compute_sku_list = None


def _get_gcp_instance_pricing(
    credentials,
    region,
    zone,
    instance_type,
    gpu_info=None
):
    global _gcp_services_list
    global _gcp_compute_sku_list

    creds = service_account.Credentials.from_service_account_info(
        json.loads(credentials)
    )
    catalog = cloud_catalog.CloudCatalogClient(credentials=creds)
    # Step one:  Find the Compute Engine service
    if not _gcp_services_list:
        _gcp_services_list = [
            x
            for x in catalog.list_services()
            if "Compute Engine" == x.display_name
        ]
    services = _gcp_services_list
    if len(services) != 1:
        raise Exception("Did not find Compute Engine Service")
    # Step two: Get all the SKUs associated with the Compute Engine service
    if not _gcp_compute_sku_list:
        _gcp_compute_sku_list = list(catalog.list_skus(parent=services[0].name))

    skus = [x for x in _gcp_compute_sku_list if region in x.service_regions]

    # To zero'th degree, pricing for an instance is made up of:
    #   # cores * Price/PerCore of instance semi-family
    #   # GB RAM * Price/GBhr of instance semi-family
    #   <OTHER THINGS - local SSD, GPUs, Tier 1 networking>  THESE ARE TODO
    #   # Disk Storage - Just assume a 20GB disk - that's what we currently get

    # Google's Billing API has SKUs, but the SKUs don't map to anything - you
    # can't get SKU info from the actual products. We have to look up sku's
    # with pricing info, and try to map the SKU's description to the actual
    # Compute infrastructure we're using.  We do have to look at the
    # "description" field, which feels hazardous and liable to change

    def price_expr_to_unit_price(expr):
        """Convert a "Price Expression" to a unit (hourly) price"""
        unit = expr.tiered_rates[0].unit_price
        return unit.units + (unit.nanos * 1e-9)

    def get_disk_price(disk_size, skus):
        def disk_sku_filter(elem):
            if elem.category.resource_family != "Storage":
                return False
            if elem.category.resource_group != "PDStandard":
                return False
            if region not in elem.service_regions:
                return False
            if not elem.description.startswith("Storage PD Capacity"):
                # Filter out 'Regional Storage PD Capacity...'
                return False
            return True

        disk_sku = [x for x in skus if disk_sku_filter(x)]
        if len(disk_sku) != 1:
            raise Exception("Failed to find singular appropriate disk")
        disk_price_expression = disk_sku[0].pricing_info[0].pricing_expression
        unit_price = price_expr_to_unit_price(disk_price_expression)
        disk_cost_per_month = disk_size * unit_price
        disk_cost_per_hr = disk_cost_per_month / (24 * 30)
        return disk_cost_per_hr

    def get_cpu_price(num_cores, instance_type, skus):
        instance_description_mapper = {
            "e2": "E2 Instance Core",
            "n2d": "N2D AMD Instance Core",
            "h3": "Compute optimized Core",
            "c3": "Compute optimized Core",
            "c4": "Compute optimized Core",
            "c2": "Compute optimized Core",
            "c2d": "C2D AMD Instance Core",
            "c3d": "C3D AMD Instance Core",
            "t2d": "T2D AMD Instance Core",
            "a2": "A2 Instance Core",
            "m1": "Memory-optimized Instance Core",  # ??
            "m2": "Memory Optimized Upgrade Premium for Memory-optimized Instance Core",  # pylint: disable=line-too-long
            "m3": "Memory-optimized Instance Core",
            "n2": "N2 Instance Core",
            "n1": "Custom Instance Core",  # ??
        }
        instance_class = instance_type.split("-")[0]
        if instance_class not in instance_description_mapper:
            raise NotImplementedError(
                "Do not yet have a price mapping for instance type "
                f"{instance_type}"
            )

        def cpu_sku_filter(elem):
            if elem.category.resource_family != "Compute":
                return False
            if elem.category.resource_group != "CPU":
                return False
            if elem.category.usage_type != "OnDemand":
                return False
            if region not in elem.service_regions:
                return False
            if "Sole Tenancy" in elem.description:
                return False
            if not elem.description.startswith(
                instance_description_mapper[instance_class]
            ):
                return False
            return True

        cpu_sku = [x for x in skus if cpu_sku_filter(x)]
        if len(cpu_sku) != 1:
            raise Exception("Failed to find singular appropriate cpu billing")
        cpu_price_expression = cpu_sku[0].pricing_info[0].pricing_expression
        unit_price = price_expr_to_unit_price(cpu_price_expression)
        cpu_price_per_hr = num_cores * unit_price
        return cpu_price_per_hr

    def get_mem_price(num_gb, instance_type, skus):
        instance_description_mapper = {
            "e2": "E2 Instance Ram",
            "n2d": "N2D AMD Instance Ram",
            "c2": "Compute optimized Ram",
            "c3": "Compute optimized Ram",
            "h3": "Compute optimized Ram",
            "c2d": "C2D AMD Instance Ram",
            "c3d": "C3D AMD Instance Ram",
            "c4": "C4 Instance RAM",
            "t2d": "T2D AMD Instance Ram",
            "a2": "A2 Instance Ram",
            "m1": "Memory-optimized Instance Ram",
            "m2": "Memory-optimized Instance Ram",
            "m3": "Memory-optimized Instance Ram",  # ??
            "n2": "N2 Instance Ram",
            "n1": "Custom Instance Ram", # ??
        }
        # TODO: Deal with 'Extended Instance Ram'
        instance_class = instance_type.split("-")[0]
        if instance_class not in instance_description_mapper:
            raise NotImplementedError(
                "Do not yet have a price mapping for instance type "
                f"{instance_type}"
            )

        def mem_sku_filter(elem):
            if elem.category.resource_family != "Compute":
                return False
            if elem.category.resource_group != "RAM":
                return False
            if elem.category.usage_type != "OnDemand":
                return False
            if region not in elem.service_regions:
                return False
            if "Sole Tenancy" in elem.description:
                return False
            if not elem.description.startswith(
                instance_description_mapper[instance_class]
            ):
                return False
            return True

        mem_sku = [x for x in skus if mem_sku_filter(x)]
        if len(mem_sku) != 1:
            raise Exception("Failed to find singular appropriate RAM billing")
        mem_price_expression = mem_sku[0].pricing_info[0].pricing_expression
        unit_price = price_expr_to_unit_price(mem_price_expression)
        ram_price_per_hr = num_gb * unit_price
        return ram_price_per_hr

    def get_accel_price(gpu_description, gpu_count, skus):
        def gpu_sku_filter(elem):
            if elem.category.resource_family != "Compute":
                return False
            if elem.category.resource_group != "GPU":
                return False
            if elem.category.usage_type != "OnDemand":
                return False
            return elem.description.lower().startswith(
                gpu_description.lower())

        gpu_sku = [x for x in skus if gpu_sku_filter(x)]

        if len(gpu_sku) != 1:
            raise Exception("Failed to find singular appropriate GPU billing")
        gpu_price_expression = gpu_sku[0].pricing_info[0].pricing_expression
        unit_price = price_expr_to_unit_price(gpu_price_expression)
        gpu_price_per_hr = gpu_count * unit_price
        return gpu_price_per_hr


    machine = _get_gcp_machine_types(credentials, zone)[instance_type]
    instance_price = (
        get_cpu_price(machine["vCPU"], instance_type, skus)
        + get_mem_price(machine["memory"] / 1024, instance_type, skus)
        # TODO: Actual disk size (20 is GHPC default)
        + get_disk_price(20.0, skus)
    )
    if gpu_info:
        (gpu_name, gpu_count) = gpu_info
        if gpu_count:
            # Need to map GPU name to GPU description for Pricing API
            try:
                gpu_desc = machine["accelerators"][gpu_name]["description"]
                instance_price += get_accel_price(gpu_desc, gpu_count, skus)
            except KeyError as err:
                raise Exception(
                    "Failed to map accelerator to instance"
                ) from err


    return instance_price


def get_instance_pricing(
    cloud_provider, credentials, region, zone, instance_type, gpu_info=None
):
    """Return price per hour for an instance"""
    if cloud_provider == "GCP":
        return _get_gcp_instance_pricing(
            credentials, region, zone, instance_type, gpu_info
        )
    else:
        raise Exception(f'Unsupported Cloud Provider "{cloud_provider}"')


def gcs_apply_bucket_acl(
    bucket, account, permission="roles/storage.objectViewer"
):

    logger.info(
        "Attempting to grant %s to gs://%s/ for user %s",
        permission,
        bucket,
        account,
    )
    client = gcs.Client()
    try:
        gcs_bucket = client.get_bucket(bucket)
        policy = gcs_bucket.get_iam_policy()
        for binding in policy.bindings:
            if binding["role"] == permission:
                binding["members"].add(account)
                break
        else:
            policy.bindings.append(
                {"role": permission, "members": set(account)}
            )

        gcs_bucket.set_iam_policy(policy)

    # Myriad errors could occur, none of them handleable so just log and move on
    except Exception as err:  # pylint: disable=broad-except
        logger.error("Failed to apply GCS Policy", exc_info=err)


def gcs_upload_file(bucket, path, contents, extra_acl=None):
    extra_acl = extra_acl if extra_acl else []
    logger.info(
        "Attempting to upload to gs://%s/%s", bucket, path if path else ""
    )
    client = gcs.Client()
    gcs_bucket = client.bucket(bucket)
    blob = gcs_bucket.blob(path)
    blob.upload_from_string(contents)
    for acl in extra_acl:
        user = acl.get("user", None)
        permission = acl.get("permission", None)
        if user and permission:
            if permission in ["OWNER", "READER", "WRITER"]:
                blob.acl.user(user).grant(permission)
        blob.acl.save()
    client.close()


def gcs_fetch_file(bucket, paths):
    client = gcs.Client()
    gcs_bucket = client.bucket(bucket)
    results = {}
    for path in paths:
        try:
            logger.debug(
                "Attempting to download from gs://%s/%s",
                bucket,
                path if path else "",
            )
            blob = gcs_bucket.blob(path)
            results[path] = blob.download_as_text()
        except google.cloud.exceptions.NotFound as nf:
            logger.info(
                "Attempt failed (Not Found) to download {path}", exc_info=nf
            )
    client.close()
    return results


def gcs_get_blob(bucket, path):
    """Returns a blob object - it may or may not exist"""
    client = gcs.Client()
    gcs_bucket = client.bucket(bucket)
    return gcs_bucket.blob(path)


def get_gcp_workbench_region_zone_info(
    credentials, service="notebooks", api_version="v1"
):

    (project, nb) = _get_gcp_client(credentials, service, api_version)
    request = nb.projects().locations().list(name=f"projects/{project}")
    result = request.execute()
    locations = [x["locationId"] for x in result["locations"]]
    return locations


def get_gcp_filestores(credentials):
    """Returns an array of Filestore instance information
    E.g.
    [
      {'createTime': ...,
       'fileShares': [{'capacityGb': '2660', 'name': 'data'}],
       'name': 'projects/<project>/locations/<zone>/instances/<name>',
       'networks': [
        {'ipAddresses': ['10.241.201.242'],
         'modes': ['MODE_IPV4'],
         'network': '<network-name>',
         'reservedIpRange': '10.241.201.240/29'
         }
       ],
       'state': 'READY',
       'tier': 'PREMIUM'
      },
      ...
    ]
    """
    (project, client) = _get_gcp_client(credentials, "file", "v1")
    request = (
        client.projects()
        .locations()
        .instances()
        .list(parent=f"projects/{project}/locations/-")
    )
    result = request.execute()
    return result["instances"]