чт, 23 нояб. 2017 г. в 20:27, Mike Bayer <mike...@zzzcomputing.com>:
> On Thu, Nov 23, 2017 at 8:44 AM, Антонио Антуан <a.ch....@gmail.com> > wrote: > > > >> A Query can have lots of entities in it, and if you're doing sharding a > >> single result set can refer to any number of shard identifiers within > >> not just a single result set but within a single row; they might have > >> come from dozens of different databases at once > > > > In my case it is not possible: all entities in query can be gotten only > from > > one particular shard. We have totally the same database structure for > each > > shard. The difference is just data stored into database. No `shard_id` or > > any other key as part of primary key for any table. > > > so just to note, these aren't "shards", they're tenants. you have a > multi-tenant application, which is normally a really easy thing. but > you have a few side applications that want to "cheat" and use the > per-tenant object model across multiple tenants simultaneously in the > scope of a single Session. > > > If I want to make query > > for particular database I always want to retrieve data ONLY from that > > database. And even more than that: ONLY one database during one session > > transaction (or, in other words, one http-request to our app). > > if you have one "tenant id" per HTTP request, the standard HTTP > request pattern is one Session per request. There's no problem in > that case. You mentioned you have some non-flask applications that > want to communicate with multiple tenants in one Session. > Yes, you're right. We have some offline actions, when we want to ask each tenant about something specific. I see, that currently the most safe way is to call `commit`, `rollback`, `remove` or `expunge_all` on session instance: all this methods drops identity map. Please let me know if I'm wrong. > > > > >> This could only be suited with a very open plugin point that is > carefully > >> architected, tested, and documented and I don't have the resources to > >> envision this for a short-term use case. > > > > I've seen a lot of questions (for example in stackoverflow) "how to > manage > > several binds with sqlalchemy" and all answers are "Use separated > sessions". > > it's probably the best answer feel free to show specifics and I can > determine if their request would fit this hypothetical feature > otherwise. > > > I really not understand, what is the problem to implement "multibound" > > session. > > The Session has always supported multiple binds. There are two > levels supported. One is per table/mapper: > > http://docs.sqlalchemy.org/en/latest/orm/persistence_techniques.html#simple-vertical-partitioning > > you could probably adapt your multiple tenants into individual > mappings if there are a limited number, see the approach at > https://bitbucket.org/zzzeek/sqlalchemy/wiki/UsageRecipes/EntityName. > > The other level is per primary key, that is, each primary key in the > identity map has a different bind. that's horizontal sharding. > > You're looking for a new level, which is, multiple binds for *one* > primary key. This is an intricate feature request that is feasible > but not in the short term. > > > > Don't have ideal vocabulary to explain how our team like it :) > > Currently I see only one problem: loading instances. Of course, after > fixing > > other problems may appear... > > > > I can (and want) make it as part of SQLAlchemy library. Fully-tested > part, > > of course. If you say that it is bad idea, ok then. > > it's not a bad idea. It's just difficult, and I can't do it right now. > > > >I can make it as a > > plugin, but there is a problem: functions in `loading` module are > monolithic > > and it needs some refactor for the plugin. May I suggest refactor as pull > > request? > > you can do a pull request but note that the PR process for SQLAlchemy > is not quick. 90% of code-related pull requests I get have no tests, > no documentation, or anything. More elaborate feature requests > typically involve that I end up doing the whole thing myself in any > case, using the submitter's original code as just a sketch, which > means that more involved PRs are usually just another form of feature > request. These PRs are almost always for Core level features as the > Core is easier for outside contributors to work on. ORM-level > contributions are extremely rare these days but of course I welcome > contributors for the ORM. > > And if so: could it be merged not only for major release but for, > > at least, 1.0.* (yes, in our project we use 1.0.19 :) )? > > The 1.0 series is in "maintenance" mode and as soon as 1.2 is released > (which is hopefully by end of year) it will go into "Security" mode. > There are no more 1.0 releases scheduled. It is not reasonable to be > doing new SQLAlchemy-oriented development without first upgrading your > application to the latest release which in this case is 1.1.15. > > > > I really don't want (of course!) to copy entire `loading` module for > > additional logic of `identitykey` construction. But currently I do not > see > > any other way to implement it for my project :( > > > >> when you query two different databases, you are using > >> two independent transactions in any case; > > > > So, what is the difference, if there are two transactions in any case? :) > > because you would have two identity maps > > > > >> I don't understand why you can't use independent sessions > > > > The first problem is `query` property of `Base` instances. If we use > several > > sessions, we need to use the same amount of `Base` classes and, > > "Base" doesn't have a "query" property - that's an anti-pattern > popularized by flask-sqlalchemy, for which SQLAlchemy's > "scoped_session" also has marginal support for, because people insist > upon it and it is very simple to implement, however, it's a bad idea. > I would lose that pattern ASAP. That simple anti-pattern is > causing you to have to write a whole new elaborate Session and Query > and get new features implemented in the deepest parts of the ORM. > > But beyond that, you can even keep Base.query and just use a > scoped_session() that also selects for the shard_id you care about, or > use your custom Query subclass to switch the Session transparently. > > To sum up: 1. Base.query is a bad idea and 2. you can still use it > with multiple sessions anyway as you're willing to subclass Query in > any case. > > > > > Another problem for us is already existed code. We can use sessions > > registry, but it take a lot of month to override entire project. Another > > way: append into `Query.__iter__` such a code: > >>> self.with_session(sessions.get(self._shard_id)) > >>> return super(Query, self).__iter__() > > But it has no effect for UPDATE and INSERT queries. Also, I'm not sure > that > > there is no problems in that way... > > With all due respect, that is quite a request. I apologize for my > bluntness but the this is simply not how open source software works. > Your team, which I assume is a commercially supported team of > programmers paid full time to work on this application, have > architected around a pattern that is not supported by the tools you > are using - open source tools that are maintained mostly by one person > (me) and for which you are using free of charge. While i am glad to > look to eventually supporting some of your use case, it is not > reasonable that I drop everything and rearchitect key parts of my > library at once in order solve your company's problem, for which you > can instead make use of your full-time paid team of developers to > adjust on your end using one of many alternatives that have been > proposed. I have a full time job for which maintenance of > SQLAlchemy is only a small part of that time, and I apologize but I do > not have the time to implement this feature for your company right > now. > > Excuse me if I let you think that I want you to implement this. Of course I understand all your notes below. I have never admit an idea that you can leave all your affairs and run to solve my problem. All my questions is just about problems of my team and our methods to solve it, not about explicit actions. When I said about `Query.__iter__` additional code, looks like I mispronounced. Of course, I said about our `Query` (`CustomQuery`) implementation. > > > I have one more thought. Don't you think that it is some kind of bug: I > make > > query for one bind and got entity from another. Yes, that behavior is not > > foreseen by library. But from other point of view, library docs have > > examples how to use several binds within one session. So, problem may > > happens not only in my case. > > Again, I'm into the 12th year of doing this and no such request has > ever been made, so this is not a widely requested feature. Having > supplemental keys in the identity key is not a bad idea but within the > realm of the ORM, no feature is simple. Everything added to the ORM > runs the risk of becoming a liability both in terms of regressions as > well as the feature being mis-designed and then needing to be replaced > with something else. The ORM is loaded with history like this, some > of it long gone, some of it pushed onto the sidelines, and some of it > still detracting from the tool today. It's not possible to overstate > how difficult it is to add new features to the ORM that don't wind up > creating enormous efforts for years going forward. The development > process is slow and methodical and the whole community benefits > greatly from this conservatism. > Conservatism, got it :) Ok, I want to try to imlement this feature by myself. I hope that my efforts were crowned with success and, more than that, appear in your library, sooner or later. Will be great if you continue to help me with it. I see several key places, that should be tested in that case: loading module and its funtions `load_on_ident`, `load_scalar_attributes`, `get_from_identity` and `instances`... So, actually, entire module :) Is there any other key places in your library that I should note in my case? > > > > Anyway, can my suggestion > > (https://gist.github.com/aCLr/746f92dedb4d303a49033c0db22beced) has any > > effect for classic one-bound `Session`? If it can't, so, what's the > problem? > > :) > > > > > > Excuse me for wasting your time. > > And excuse me if my suggestions are idiotic :) > > > > Appreciate your help. > > > > чт, 23 нояб. 2017 г. в 0:20, Mike Bayer <mike...@zzzcomputing.com>: > >> > >> On Wed, Nov 22, 2017 at 4:56 AM, Антонио Антуан <a.ch....@gmail.com> > >> wrote: > >> > Glad to see that you remember my messages :) > >> > > >> > I've dived into `loading` module and I see that currently it is really > >> > complicated to store additional data into pkey for each instance. > >> > > >> > Unfortunately, suggested solutions not good for my project. > >> > Also, I think that `shard` meaning in my case is not the same as > usual. > >> > > >> > I want to describe structure of out project, maybe it can help. > >> > > >> > Here is definition of our databases structure: > >> > http://joxi.ru/nAyJVvGiXMv0Dr. > >> > We got master db and several geo databases. Catalogs like `users`, > >> > `groups`, > >> > `offers` and other are replicating to geo databases, so that data is > >> > always > >> > the same. > >> > But also we have tables like `clicks` and `leads`. Each app instance > >> > contains the data about them in database, related to its geo: > >> > europe-instance into europe-db, usa-instance into usa-database and so > >> > on. > >> > Periodically master-app pulls clicks and leads to master-database. > >> > Synced > >> > objects always have different ids into master- and get-db, so it is > ok. > >> > > >> > But one time project owner came and said: "I need SAAS". > >> > We see, that in current structure it's very hard (and really ugly) to > >> > implement saas-solution. Amount of `Base*`, `Session*`, `Order*` and > >> > other > >> > models will be multiplied with tenants amount. > >> > > >> > I discovered that I can override `get_bind` with another logic and it > >> > was > >> > great: we can remove several `Base` classes, several Sessions and > >> > several > >> > `Orders`. > >> > > >> > Mechanism looks like this: > >> > - we use one instance on each geo for all tenants. > >> > - we create separated databases for each tenant: this will be > multiplied > >> > with tenants amount: http://joxi.ru/nAyJVvGiXMv0Dr. > >> > - we detect `tenant_id` using `request.host` (we use flask): each > domain > >> > binds with particular tenant; > >> > - we store `tenant_id` into global storage. > >> > - we use stored `tenant_id` into `Session.get_bind`: > >> > https://gist.github.com/aCLr/9f578a2eeb225e7d65099ffb49aa8f3a > >> > - into flask `teardown_request` we clear `tenant_id` storage and call > >> > `Session.remove()` > >> > - if we need to read from another get, just write `query = > >> > query.set_shard_id(GEO)` `tenant_id` > >> > > >> > For celery we use this: > >> > https://gist.github.com/aCLr/d8c5ac38956947da092375b2f89d7b50 > >> > Clear to. > >> > > >> > All this leads us only to pros, without cons: any developer has no > need > >> > to > >> > think about database chosing, just write code like there is only one > >> > database. If you need to read from another geo-database, just call > >> > `query.set_shard(GEO)`, tenant will be appended automatically to it. > >> > > >> > Problems begin when we tried to test non-flask and non-celery scripts, > >> > like > >> > cron tasks: we may want to query several tenant-databases during one > >> > SQLA-transaction, somethins like in my first example: > >> > https://gist.github.com/aCLr/ff9462b634031ee6bccbead8d913c41f > >> > (`assert_got_correct_objects_with_remove` and > >> > `assert_got_cached_objects_without_remove`). The result you know. > >> > > >> > > >> > During writing this message, I found out, that we need only one > >> > additional > >> > data for primary key: `connection.bind.url`. I see, that SQLA already > >> > have > >> > it inside `_instance_processor`, so it exists inside `_instance`. I > >> > think, > >> > that `identity_key` should be constructed (in my case) with this code: > >> > https://gist.github.com/aCLr/746f92dedb4d303a49033c0db22beced. Clear, > >> > don't > >> > you think so? > >> > >> that's where something needs to happen but SQLAlchemy can't do this in > >> such a way that is hardcoded to exactly your particular use case. A > >> Query can have lots of entities in it, and if you're doing sharding a > >> single result set can refer to any number of shard identifiers within > >> not just a single result set but within a single row; they might have > >> come from dozens of different databases at once. This could only be > >> suited with a very open plugin point that is carefully architected, > >> tested, and documented and I don't have the resources to envision this > >> for a short-term use case. > >> > >> > Problems begin when we tried to test non-flask and non-celery scripts, > >> > like > >> > cron tasks: we may want to query several tenant-databases during one > >> > SQLA-transaction, > >> > >> I don't understand why you can't use independent sessions for this, > >> because when you query two different databases, you are using two > >> independent transactions in any case; they are only coordinated if > >> one is using two-phase transactions which I doubt is the case here > >> (while SQLAlchemy put a lot of work into making that possible, I don't > >> think anyone has ever used that feature). > >> > >> Instead of: > >> > >> Session.query(Foo).set_shard(id) > >> > >> you say: > >> > >> sessions.get(shard_id).query(Foo) > >> > >> > >> > >> > >> > >> > > >> > > >> > > >> > вт, 21 нояб. 2017 г. в 19:15, Mike Bayer <mike...@zzzcomputing.com>: > >> >> > >> >> I've looked to see how hard it would be to allow "supplemental" > >> >> attributes to form part of the mapper's primary key tuple, and it > >> >> would be pretty hard. The "easy" part is getting the mapper to set > >> >> itself up with some extra attributes that can deliver some kind of > >> >> supplemental value to the identity key. the harder part is then in > >> >> loading.py where we get new rows from the DB and need this > >> >> value...which means some whole new kind of system would need to > >> >> deliver this for any arbitrary part of the result set given a mapping > >> >> and the selectable we're looking at (keep in mind a Query can have > >> >> lots of the same mapping in a single row with aliases). This would > >> >> be very complicated to implement and test. I am not seeing any > quick > >> >> way to suit this use case, which has also not ever been requested > >> >> before. > >> >> > >> >> > >> >> > >> >> On Tue, Nov 21, 2017 at 10:12 AM, Mike Bayer < > mike...@zzzcomputing.com> > >> >> wrote: > >> >> > On Tue, Nov 21, 2017 at 7:39 AM, Антонио Антуан < > a.ch....@gmail.com> > >> >> > wrote: > >> >> >> Hi guys. > >> >> >> > >> >> >> I got this code example: > >> >> >> https://gist.github.com/aCLr/ff9462b634031ee6bccbead8d913c41f. > >> >> >> > >> >> >> Here I make custom `Session` and custom `Query`. As you see, > >> >> >> `Session` > >> >> >> has > >> >> >> several binds. > >> >> >> > >> >> >> Also, you can see that there are two functions: > >> >> >> `assert_got_correct_objects_with_remove` and > >> >> >> `assert_got_cached_objects_without_remove`. > >> >> >> > >> >> >> The first checks that we got correct results if `Session.remove` > >> >> >> called. > >> >> >> The second checks, that we got incorrect results if > `Session.remove` > >> >> >> not > >> >> >> called. > >> >> >> > >> >> >> I understand, that behavior is correct: we don't remove session - > >> >> >> so, > >> >> >> we got > >> >> >> same result from "cache-like" > >> >> >> `sqlalchemy.orm.loading._instance_processor.session_identity_map`. > >> >> >> > >> >> >> I want to avoid that mechanism and don't want to use > >> >> >> `session_identity_map` > >> >> >> for different binds. In ideal, bind should be used as part of key > >> >> >> for > >> >> >> `session_identity_map`, but I guess, that it is not possible. > >> >> >> Another way, acceptable for me: disable this mechanism. But I do > not > >> >> >> found > >> >> >> ways to achieve this. > >> >> >> And the third option: construct instances manually. Looks like I > >> >> >> should > >> >> >> copy > >> >> >> code from `loading` module and add that method to `CustomSession`: > >> >> > > >> >> > > >> >> > there's really no reason at all to use a "ShardedSession" if you > have > >> >> > overlapping primary key spaces from each of your binds. I'm not > >> >> > sure > >> >> > if I mentioned this at the beginning of the emails regarding this > >> >> > project but I hope that I mentioned just using separate Session > >> >> > objects is vastly simpler for non-intricate sharding cases, such as > >> >> > where you always know which shard you care about and you don't care > >> >> > about any of the others for a certain operation. The point of > >> >> > ShardedSession is so that objects pulled from multiple databases > can > >> >> > be intermingled in the same query and in the same transaction - > which > >> >> > by definition means they have unique primary keys. If that's not > >> >> > what you're doing here I don't see what advantage all this > complexity > >> >> > is getting you. > >> >> > > >> >> > If you're still convinced you need to be using a monolithic > >> >> > ShardedSession then there needs to be some kind of translation of > >> >> > data > >> >> > such that the mapper sees unique primary keys across the shards, or > >> >> > unique classes. > >> >> > > >> >> > I've tried to think of ways to do this without too much difficulty > >> >> > but > >> >> > none of them are really worth the complexity and hackiness it would > >> >> > require. The absolutely quickest and most well-supported, no > hacks > >> >> > required way would be to properly create a composite primary key on > >> >> > your classes, where the second column is your shard id: > >> >> > > >> >> > class A(Base): > >> >> > __tablename__ = 'a' > >> >> > > >> >> > id = Column(Integer, primary_key=True) > >> >> > shard_id = Column(Integer, primary_key=True) > >> >> > > >> >> > I tried to see if the "shard_id" column can be some kind of > >> >> > expression > >> >> > that is not a Column on the Table but the mapper() is not set up to > >> >> > support this unless you mapped the whole class to a select() > >> >> > construct, which would make for too-complicated SQL, and you'd > still > >> >> > need to intercept this select() using events to put the right shard > >> >> > id > >> >> > in. Another is to create a custom column that renders in a special > >> >> > way, but again you need to create events to intercept it in every > >> >> > case > >> >> > to put the right shard id in, and/or remove it from things like > >> >> > insert() statements. > >> >> > > >> >> > by far your two best solutions are: 1. use separate Session objects > >> >> > per shard 2. make sure your data actually has shard-specific > primary > >> >> > keys > >> >> > > >> >> > > >> >> > > >> >> > > >> >> > > >> >> > > >> >> > > >> >> >> > >> >> >> def instances(self, cursor, __context=None): > >> >> >> context = __context > >> >> >> if context is None: > >> >> >> context = QueryContext(self) > >> >> >> return self._custom_instances(self, cursor, context) > >> >> >> > >> >> >> > >> >> >> > >> >> >> def custom_instances(query, cursor, context): > >> >> >> """copied from `loading.instances` code with disabled > >> >> >> `session_identity_map`""" > >> >> >> > >> >> >> > >> >> >> > >> >> >> The third way is the most ugly and I want to avoid it. > >> >> >> > >> >> >> Could you help me with my hard choice and, maybe, suggest any > other > >> >> >> ways and > >> >> >> options? :) > >> >> >> > >> >> >> Thank you. > >> >> >> > >> >> >> -- > >> >> >> SQLAlchemy - > >> >> >> The Python SQL Toolkit and Object Relational Mapper > >> >> >> > >> >> >> http://www.sqlalchemy.org/ > >> >> >> > >> >> >> To post example code, please provide an MCVE: Minimal, Complete, > and > >> >> >> Verifiable Example. See http://stackoverflow.com/help/mcve for a > >> >> >> full > >> >> >> description. > >> >> >> --- > >> >> >> You received this message because you are subscribed to the Google > >> >> >> Groups > >> >> >> "sqlalchemy" group. > >> >> >> To unsubscribe from this group and stop receiving emails from it, > >> >> >> send > >> >> >> an > >> >> >> email to sqlalchemy+unsubscr...@googlegroups.com. > >> >> >> To post to this group, send email to sqlalchemy@googlegroups.com. > >> >> >> Visit this group at https://groups.google.com/group/sqlalchemy. > >> >> >> For more options, visit https://groups.google.com/d/optout. > >> >> > >> >> -- > >> >> SQLAlchemy - > >> >> The Python SQL Toolkit and Object Relational Mapper > >> >> > >> >> http://www.sqlalchemy.org/ > >> >> > >> >> To post example code, please provide an MCVE: Minimal, Complete, and > >> >> Verifiable Example. See http://stackoverflow.com/help/mcve for a > full > >> >> description. > >> >> --- > >> >> You received this message because you are subscribed to the Google > >> >> Groups > >> >> "sqlalchemy" group. > >> >> To unsubscribe from this group and stop receiving emails from it, > send > >> >> an > >> >> email to sqlalchemy+unsubscr...@googlegroups.com. > >> >> To post to this group, send email to sqlalchemy@googlegroups.com. > >> >> Visit this group at https://groups.google.com/group/sqlalchemy. > >> >> For more options, visit https://groups.google.com/d/optout. > >> > > >> > -- > >> > > >> > Антон > >> > > >> > -- > >> > SQLAlchemy - > >> > The Python SQL Toolkit and Object Relational Mapper > >> > > >> > http://www.sqlalchemy.org/ > >> > > >> > To post example code, please provide an MCVE: Minimal, Complete, and > >> > Verifiable Example. See http://stackoverflow.com/help/mcve for a full > >> > description. > >> > --- > >> > You received this message because you are subscribed to the Google > >> > Groups > >> > "sqlalchemy" group. > >> > To unsubscribe from this group and stop receiving emails from it, send > >> > an > >> > email to sqlalchemy+unsubscr...@googlegroups.com. > >> > To post to this group, send email to sqlalchemy@googlegroups.com. > >> > Visit this group at https://groups.google.com/group/sqlalchemy. > >> > For more options, visit https://groups.google.com/d/optout. > >> > >> -- > >> SQLAlchemy - > >> The Python SQL Toolkit and Object Relational Mapper > >> > >> http://www.sqlalchemy.org/ > >> > >> To post example code, please provide an MCVE: Minimal, Complete, and > >> Verifiable Example. See http://stackoverflow.com/help/mcve for a full > >> description. > >> --- > >> You received this message because you are subscribed to the Google > Groups > >> "sqlalchemy" group. > >> To unsubscribe from this group and stop receiving emails from it, send > an > >> email to sqlalchemy+unsubscr...@googlegroups.com. > >> To post to this group, send email to sqlalchemy@googlegroups.com. > >> Visit this group at https://groups.google.com/group/sqlalchemy. > >> For more options, visit https://groups.google.com/d/optout. > > > > -- > > > > Антон > > > > -- > > SQLAlchemy - > > The Python SQL Toolkit and Object Relational Mapper > > > > http://www.sqlalchemy.org/ > > > > To post example code, please provide an MCVE: Minimal, Complete, and > > Verifiable Example. See http://stackoverflow.com/help/mcve for a full > > description. > > --- > > You received this message because you are subscribed to the Google Groups > > "sqlalchemy" group. > > To unsubscribe from this group and stop receiving emails from it, send an > > email to sqlalchemy+unsubscr...@googlegroups.com. > > To post to this group, send email to sqlalchemy@googlegroups.com. > > Visit this group at https://groups.google.com/group/sqlalchemy. > > For more options, visit https://groups.google.com/d/optout. > > -- > SQLAlchemy - > The Python SQL Toolkit and Object Relational Mapper > > http://www.sqlalchemy.org/ > > To post example code, please provide an MCVE: Minimal, Complete, and > Verifiable Example. See http://stackoverflow.com/help/mcve for a full > description. > --- > You received this message because you are subscribed to the Google Groups > "sqlalchemy" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to sqlalchemy+unsubscr...@googlegroups.com. > To post to this group, send email to sqlalchemy@googlegroups.com. > Visit this group at https://groups.google.com/group/sqlalchemy. > For more options, visit https://groups.google.com/d/optout. > -- Антон -- SQLAlchemy - The Python SQL Toolkit and Object Relational Mapper http://www.sqlalchemy.org/ To post example code, please provide an MCVE: Minimal, Complete, and Verifiable Example. See http://stackoverflow.com/help/mcve for a full description. --- You received this message because you are subscribed to the Google Groups "sqlalchemy" group. To unsubscribe from this group and stop receiving emails from it, send an email to sqlalchemy+unsubscr...@googlegroups.com. To post to this group, send email to sqlalchemy@googlegroups.com. Visit this group at https://groups.google.com/group/sqlalchemy. For more options, visit https://groups.google.com/d/optout.
