Alembic：使用`version_table_schema`时'关系“public.alembic_version”不存在'

Question

我正在为 Alembic 编写一些自定义代码，以使我的数据库在我的项目开发环境中始终保持更新。该项目涉及的数据库具有以下内容：

• 用于共享数据的 public 架构
• 每个客户端“数据库”一个模式
• 一个架构充当所有客户端架构（组织）的prototype

此时，我并不担心多个客户端架构，只需要保持 public 和 prototype 架构是最新的。我的 env.py 脚本适用于 public 架构，但不适用于 prototype，因为在使用 prototype 时，alembic 正在尝试使用来自 public 的版本表。

所以，我想我可以使用version_table_schema 选项来维护public 架构中的一个版本表和prototype 架构中的一个版本表。但是，一旦我开始使用它，当我尝试进行升级时，就会收到“relation "public.alembic_version" 不存在”错误。

我看到的唯一区别是，当我使用设置为适当架构的version_table_schema 时，生成的修订脚本实际上包含指向op.drop_table('alembic_version') 的行。该行仅在使用version_table_schema 时存在。

我希望我只是错过了一些小事。

以下是所有相关的源文件供参考。唯一的外部依赖应该是appcontext 和getDeclarativeBase。 appcontext 仅用于配置，并且肯定可以正常工作（数据库连接正常）。 getDeclarativeBase 是一种动态获取模式的声明性基础（和相关元数据）的方法。根据调试输出，这似乎也可以正常工作。元数据对象本身在构造时与正确的模式相关联。

---

用于运行迁移的包装函数。autoMigratePublic() 和 autoMigrateOrgProto() 在这种情况下是有问题的方法

""" A wrapper around Alembic that will assist with database migrations.

    Our database system is soemwhat complex to upgrade. There is a public
    schema which we use for shared things (such as authentication and job
    management). Then there is the org prototype.

    There are also individual schemas for each client org. These are trickier.

    http://stackoverflow.com/a/35211383/703040

    Also useful:

    https://github.com/zzzeek/alembic/blob/0e3319bb36f1612e41a8d7da5a48ce1ca33a0b2b/alembic/config.py#L476
"""

import logging
import os
import time

import alembic.config
import alembic.util

CONFIG_ORG_PROTO = os.path.join("migrations", "alembic_orgproto.ini")
CONFIG_PUBLIC = os.path.join("migrations", "alembic_public.ini")

log = logging.getLogger("sys.migrations")

def autoMigratePublic():
    try:
        freezePublic()
    except alembic.util.CommandError:
        log.warn("[public] Problem creating migration revision. Skipping as this sometimes just means that we are working with a new DB.")
    upgradePublic()
    return

def autoMigrateOrgProto():
    try:
        freezeOrgProto()
    except alembic.util.CommandError:
        log.warn("[orgproto] Problem creating migration revision. Skipping as this sometimes just means that we are working with a new DB.")
    upgradeOrgProto()
    return

def freezePublic():
    log.info("[public] Checking the database for revisions...")
    alembicArgs = [
        "--config", CONFIG_PUBLIC,
        "--raiseerr",
        "revision",
        "--autogenerate",
        "--message", "autogenerate {0}".format(makeRevisionName()),
    ]
    runAlembic(alembicArgs)
    return

def freezeOrgProto():
    log.info("[orgproto] Checking the database for revisions...")
    alembicArgs = [
        "--config", CONFIG_ORG_PROTO,
        "--raiseerr",
        "revision",
        "--autogenerate",
        "--message", "autogenerate {0}".format(makeRevisionName()),
    ]
    runAlembic(alembicArgs)
    return

def makeRevisionName():
    return time.strftime('%Y-%m-%d %H:%M:%S')

def upgradePublic():
    log.info("[public] Performing database upgrade...")
    alembicArgs = [
        "--config", CONFIG_PUBLIC,
        "--raiseerr",
        "upgrade",
        "head",
    ]
    runAlembic(alembicArgs)
    return

def upgradeOrgProto():
    log.info("[orgproto] Performing database upgrade...")
    alembicArgs = [
        "--config", CONFIG_ORG_PROTO,
        "--raiseerr",
        "upgrade",
        "head",
    ]
    runAlembic(alembicArgs)
    return

def upgradeOrg(schemaName):
    log.info("[%s] Performing database upgrade...", schemaName)
    alembicArgs = [
        "--config", CONFIG_ORG_PROTO,
        "--raiseerr",
        "upgrade",
        "head",
        "-x", "schema={0}".format(schemaName),
    ]
    runAlembic(alembicArgs)
    return

def runAlembic(args):
    return alembic.config.main(args)

---

执行迁移的类。这个类在 env.py 文件中被引用

import copy
import logging
import os
import re
import traceback

from logging.config import fileConfig

from sqlalchemy import create_engine

import core.context.appcontext
from core.database.declarative import getDeclarativeBase

logging.getLogger("alembic").setLevel(logging.DEBUG)

#==============================================================================
class Migrator(object):

    def __init__(self):
        from alembic import context
        self.context = context
        self.config = context.config
        self.log = logging.getLogger("sys.migrations")
        self.sys = core.context.appcontext.instance()
        self.schema = self.config.get_main_option("schema")
        if self.context.get_x_argument("schema"):
            self.schema = self.context.get_x_argument("schema")
        return

    def start(self):
        import core.database.tables  # Make sure the metadata is defined
        if self.context.is_offline_mode():
            self.log.error("[%s] Can't run migrations offline", self.schema)
            return
        self.doMigration()
        return

    def doMigration(self):
        targetMetadata = getDeclarativeBase(self.schema).metadata
        engine = create_engine(self.sys.getConnectionUrl(), echo=False)
        self.log.info("[%s] Engine:   %s", self.schema, engine)
        self.log.debug("[%s] Metadata: %s", self.schema, targetMetadata)
        for t in targetMetadata.sorted_tables:
            self.log.debug("    - %s", t)
        conn = engine.connect()
        try:
            self.context.configure(
                conn,
                version_table_schema=self.schema,
                target_metadata=targetMetadata,
                process_revision_directives=self.process_revision_directives,
            )
            with self.context.begin_transaction():
                self.context.run_migrations()
        finally:
            conn.close()
        return

    def process_revision_directives(self, context, revision, directives):
        """ Used to prevent creating empty migrations

            See: http://alembic.readthedocs.org/en/latest/cookbook.html#don-t-generate-empty-migrations-with-autogenerate
        """
        if self.config.cmd_opts.autogenerate:
            script = directives[0]
            if script.upgrade_ops.is_empty():
                self.log.debug("[%s] Auto-generated migration is empty. No migration file will be created.", self.schema)
                directives[:] = []
            else:
                self.log.info("[%s] Creating new auto-generated migration revision.", self.schema)
        return

---

样例env.py。公开升级和原型升级内容相同

from migrations.migrator import Migrator
Migrator().start() 

---

示例配置 ini 公开迁移和原型迁移使用几乎完全相同的文件

# A generic, single database configuration.

[alembic]
# path to migration scripts
script_location = migrations/orgproto

# template used to generate migration files
# file_template = %%(rev)s_%%(slug)s

# max length of characters to apply to the
# "slug" field
#truncate_slug_length = 40

# set to 'true' to run the environment during
# the 'revision' command, regardless of autogenerate
# revision_environment = false

# set to 'true' to allow .pyc and .pyo files without
# a source .py file to be detected as revisions in the
# versions/ directory
# sourceless = false

# version location specification; this defaults
# to alembic/versions.  When using multiple version
# directories, initial revisions must be specified with --version-path
# version_locations = %(here)s/bar %(here)s/bat alembic/versions

# the output encoding used when revision files
# are written from script.py.mako
# output_encoding = utf-8

sqlalchemy.url = <Not Loaded>

schema = orgproto

---

Mako 文件我使用的是默认值。

"""${message}

Revision ID: ${up_revision}
Revises: ${down_revision | comma,n}
Create Date: ${create_date}

"""

# revision identifiers, used by Alembic.
revision = ${repr(up_revision)}
down_revision = ${repr(down_revision)}
branch_labels = ${repr(branch_labels)}
depends_on = ${repr(depends_on)}

from alembic import op
import sqlalchemy as sa
${imports if imports else ""}

def upgrade():
    ${upgrades if upgrades else "pass"}


def downgrade():
    ${downgrades if downgrades else "pass"}

Answer 1

我收到了解决问题的sqlalchemy-alembic google group 回复。在此处发布以防对其他人有所帮助：

您可能会遇到非常混乱的架构规则，这些规则适用于 PostgreSQL。看 http://docs.sqlalchemy.org/en/rel_1_0/dialects/postgresql.html#remote-schema-table-introspection-and-postgresql-search-path 详情。简短的回答是“空白”的模式和模式 “public”在 Python 端是两个不同的东西，导致很多 的混乱。

为了说服 autogenerate 根本不影响 alembic_version 无论它在哪里弹出，您都可能需要创建一个排除项 使用 include_object 的规则： http://alembic.readthedocs.org/en/latest/api/runtime.html?highlight=include_object#alembic.runtime.environment.EnvironmentContext.configure.params.include_object

def include_object(object, name, type_, reflected, compare_to): 
    if (type_ == "table" and name == 'alembic_version'): 
        return False 
    else: 
        return True

Answer 2

在我们的例子中，帮助将 alembic_version 添加到 migrations/alembic.ini 的 [alembic:exclude] 部分：

[alembic:exclude]
- tables = spatial_ref_sys,topology,layer
+ tables = spatial_ref_sys,topology,layer,alembic_version

接下来是migrations/env.py:

context.configure(
    version_table_schema='public',
    # other params
)