I have a rather odd situation here.
My OrdersController has an edit method and at some point I have the following statement:
order_item.item = item
As you can guess, order_item.item expects a reference to an Item object, which is exactly what item is. Here comes the mistery. The first time I send a request to orders/{some_order_id}/edit, I receive the expected response. But if I send a new request (even if I just press F5) the application crashes and I receive the following error message:
ActiveRecord::AssociationTypeMismatch in OrdersController#edit
Item(#177601092) expected, got Item(#67520280)
What have I discovered so far? I have put an puts "#{Item.object_id}" just before the order_item.item = item and found out that Item.object_id changes from a request to another, which means that the Item ActiveRecord is changing its position on memory. (For instance, in the above example, during the first request Item.object_id is 67520280, while in the second it changes to 177601092.)
My question is: What can be causing the Item ActiveRecord to change its position on memory?
Useful info: I'm using rails 3.0.20
Update:
I found out that this change of ID is pretty common. But I still can't figure out why the expectation doesn't change as the Item.object_id does.
Despite the fact that I am not using Factory Girl, this answer and this discussion helped me a lot.
tl;dr
The problem was solved when I set cached_classes to true in my environment.
Related
I am trying to implement 2FA(two factor authentication) in my existing rails 4.2.10 application, I have configured many bits.
Issue I am facing is to get/retrieve a code which is valid for 5 minutes and send this code over to user on his defined phone number or email.
I did tried ROTP::TOTP.new(user.otp_secret).at(Time.now), guessing from gem's source code, which seems to work fine and give a valid otp_code in console, but in sessions_controler, as weird as it sounds, user.otp_secret is null, always...
I have posted an issue on the gem.
I don't think this can be bug, rather this is a functionality I want to build.
My stack:
Ruby: 2.4.2
Rails: 4.2.10
Devise: 4
attr_encrypted: 1.4(if it matters)
Additionally, I want to extend drift period(code acceptance time) to 5 minutes. I think that will be easy, but doing it for single code, not universally, or for all codes, this has me thinking for a while now.
My main issue is the first one, getting the code to send through SMS, this is a subproblem, which I think is doable, but if anyone has/had experience with this and can help, that will be great.
UPDATE: I updated attr_encrypted and restarted the system, it started working, also I realized there is a method current_otp in which devise_two_factor adds in the user model, so I started using that. BUT after a few minutes, it is also throwing the same issue of user.otp_secret being nil. Its getting weird...
UPDATE 2/Hacky solution: Weirdly enough, I had to add these 3 methods in user model and everything started working:
def encrypted_otp_secret
self[:encrypted_otp_secret]
end
def encrypted_otp_secret_iv
self[:encrypted_otp_secret_iv]
end
def encrypted_otp_secret_salt
self[:encrypted_otp_secret_salt]
end
As you can suspect, i got here by examining a behavior thatdoing user.encrypted_otp_secret was giving me nil while it was not, even after reloading user model. And doing user[:encrypted_otp_secret] was giving me the actual value.
It seems like a bug in attr_encrypted. I am not sure yet.
For anyone else that runs into this issue, I have found a next step needed to get the current_otp method to work. In the method pre_otp method call
> u = User.find_by(email: 'test#example.com')
> u.otp_required_for_login = true
> u.otp_secret = User.generate_otp_secret
> u.save!
and then you can call u.current_otp...
https://blog.tommyku.com/blog/integrating-two-step-two-factor-authentication-into-rails-4-project-with-devise/
As the title says, I have a simple ActiveResource in my application that is supposed to get data from an api. The collection works perfectly both locally and on the production server. However, .find doesn't work in production, i get a weird error:
MyResource.find(1, params: { website_id: 2 })
ArgumentError: wrong number of arguments (given 2, expected 0..1)
The same query works if I run it from local console connected to the api. API returns a valid response. I have no idea how else to debug it.
class MyResource < ActiveResource::Base
self.site = Rails.configuration.content_url
self.prefix = "/api/websites/:website_id/"
self.element_name = "game"
end
I have to say, i have other resources in the application using the same api, find method works for them, only this one has issues.
for some reason it makes a call to
https://github.com/rails/rails/blob/master/activerecord/lib/active_record/core.rb#L330
with 2 arguments, but i can't figure out who makes such call. And it only happens in production mode, not in dev.
I found the problem, but it's the weirdest thing ever.
In my show method in the api controller i was returning an object but i was merging the hash with some extra data. If i opened in the browser, the response was fine, development worked, but for some reason, in production it doesn't want to accept the merge.
So problem solved, but still have no idea why it's happening like that.
I ran into a similar issue in an app that used ActiveRecord for a vendor model and ActiveResource for a delivery model that has a vendor attribute. In development, if the vendor model has not yet been loaded, you can fetch a delivery and deserialize it, including the vendor attribute. If, however, the vendor model has been loaded already, attempting to fetch a delivery leads to the error described in this issue.
How to reproduce in a console:
Vendor.connection
Delivery.find(1)
Seems that AResource is trying to create a new instance of Vendor.
I think I've eliminated everything, but I'm not sure I understand OL perfectly enough to be sure. In general, let's say you and I are on a team to keep a foo up to date. I'm in one room and decide to save time I'll update the foo myself. So I start updating it. A minute later you have the same idea and log onto the edit page to update it as well. What happens if I finish first? What happens if you finish first? In a configuration where it fails how does it distinguish between someone editing and someone reading. If I catch and reload to update the lock I lose all my changes, how is this solved? Here, it's simple to redo the update, but potentially it's part of a more complicated form object.
My specific problem came when (best I can make out) loaded one copy in my browser, later forgot about it and then one in my console (also lock: 0?) couldn't update the one in my console with stale object error. Noticed the browser thing. Closed my console. Tried to reload my browser and got stale object error as well. Here's the code that's failing:
=> 7: self.update_attributes({
8: failed_view_attempts: self.failed_view_attempts += 1,
9: failed_view_at: Time.now
10: })
11: end
(byebug) self
#<Product id: 12... lock_version: 0>
#=> ActiveRecord::StaleObjectError (Attempted to update a stale object: Product.)
Things I've tried:
To see if another instance was being loaded I added puts "CALLED !!!!" in an after_initializecallback, but it only printed once.
And checking self.changed after rescuing from the error and get back ["updated_at", "failed_view_attempts", "failed_view_at"]
Need to set lock_version column default to zero (0).
Optimistic locking is based on an object version number.
Reading and object should not affect the version number at all.
If you try to update some object, this version number is compared (in sql actually update statement is used for example "update ... where version = 1 and ..." ) and increased on update.
If the comparison fails when trying to update, you get the stale object error. This means that the object was modified by somebody else while you was doing your changes.
To resolve this error you need to load the object again to get the actual version and maybe merge the changes manually (presenting the user some info about it, and let user decide for example).
I have the following code in a rails model:
foo = Food.find(...)
foo.with_lock do
if bar = foo.bars.find_by_stuff(stuff)
# do something with bar
else
bar = foo.bars.create!
# do something with bar
end
end
The goal is to make sure that a Bar of the type being created is not being created twice.
Testing with_lock works at the console confirms my expectations. However, in production, it seems that in either some or all cases the lock is not working as expected, and the redundant Bar is being attempted -- so, the with_lock doesn't (always?) result in the code waiting for its turn.
What could be happening here?
update
so sorry to everyone who was saying "locking foo won't help you"!! my example initially didin't have the bar lookup. this is fixed now.
You're confused about what with_lock does. From the fine manual:
with_lock(lock = true)
Wraps the passed block in a transaction, locking the object before yielding. You pass can the SQL locking clause as argument (see lock!).
If you check what with_lock does internally, you'll see that it is little more than a thin wrapper around lock!:
lock!(lock = true)
Obtain a row lock on this record. Reloads the record to obtain the requested lock.
So with_lock is simply doing a row lock and locking foo's row.
Don't bother with all this locking nonsense. The only sane way to handle this sort of situation is to use a unique constraint in the database, no one but the database can ensure uniqueness unless you want to do absurd things like locking whole tables; then just go ahead and blindly try your INSERT or UPDATE and trap and ignore the exception that will be raised when the unique constraint is violated.
The correct way to handle this situation is actually right in the Rails docs:
http://apidock.com/rails/v4.0.2/ActiveRecord/Relation/find_or_create_by
begin
CreditAccount.find_or_create_by(user_id: user.id)
rescue ActiveRecord::RecordNotUnique
retry
end
("find_or_create_by" is not atomic, its actually a find and then a create. So replace that with your find and then create. The docs on this page describe this case exactly.)
Why don't you use a unique constraint? It's made for uniqueness
A reason why a lock wouldn't be working in a Rails app in query cache.
If you try to obtain an exclusive lock on the same row multiple times in a single request, query cached kicks in so subsequent locking queries never reach the DB itself.
The issue has been reported on Github.
I'm trying to update one of my objects in my rails app and the changes just don't stick. There are no errors, and stepping through with the debugger just reveals that it thinks everything is updating.
Anyway, here is the code in question...
qm = QuestionMembership.find(:first, :conditions => ["question_id = ? AND form_id = ?", q_id, form_id])
qm.position = x
qm.save
For reference sake, QuestionMembership has question_id, form_id, and position fields. All are integers, and have no db constraints.
That is basically my join table between Forms and Questions.
Stepping through the code, qm gets a valid object, the position of the object does get changed to the value of x, and save returns 'true'.
However, after the method exits, the object in the db is unchanged.
What am I missing?
You may not be finding the object that you think you are. Some experimenting in irb might be enlightening.
Also, as a general rule when changing only one attribute, it's better to write
qm.update_attribute(:position, x)
instead of setting and saving. Rails will then update only that column instead of the entire row. And you also get the benefit of the data being scrubbed.
Is there an after_save?
Is the correct SQL being emitted?
In development log, you can actually see the sql that is generated.
For something like this:
qm = QuestionMembership.find(:first, :conditions => ["question_id = ? AND form_id = ?", q_id, form_id])
qm.position = x
qm.save
You should see something to the effect of:
SELECT * FROM question_memberships WHERE question_id=2 AND form_id=6 LIMIT 1
UPDATE question_memberships SET position = x WHERE id = 5
Can you output what sql you are actually seeing so we can compare?
Either update the attribute or call:
qm.reload
after the qm.save
What is the result of qm.save? True or false? And what about qm.errors, does that provide anything that makes sense to you? And what does the development.log say?
I have run into this problem rather frequently. (I was about to say consistently, but I cannot, as that would imply that I would know when it was about to happen.)
While I have no solution to the underlying issue, I have found that it seems to happen to me only when I am trying to update mysql text fields. My workaround has been to set the field to do something like:
qm.position = ""
qm.save
qm.position = x
qm.save
And to answer everyone else... when I run qm.save! I get no errors. I have not tried qm.save?
When I run through my code in the rails console everything works perfectly as evidenced by re-finding the object using the same query brings the expected results.
I have the same issue when using qm.update_attribute(... as well
My workaround has gotten me limping this far, but hopefully someone on this thread will be able to help.
Try changing qm.save to qm.save! and see if you get an exception message.
Edit: What happens when you watch the log on the call to .save!? Does it generate the expected SQL?
Use ./script/console and run this script.. step by step..
see if the position field for the object is update or not when you run line 2
then hit qm.save or qm.save!... to test
see what happens. Also as mentioned by Tim .. check the logs
Check your QuestionMembership class and verify that position does not have something like
attr_readonly :position
Best way to debug this is to do
tail -f log/development.log
And then open another console and do the code executing the save statement. Verify that the actual SQL Update statement is executed.
Check to make sure your database settings are correct. If you're working with multiple databases (or haven't changed the default sqlite3 database to MySQL) you may be working with the wrong database.
Run the commands in ./script/console to see if you see the same behavior.
Verify that a similar object (say a Form or Question) saves.
If the Form or Question saves, find the difference between the QuestionMembership and Form or Question object.
Turns out that it was emitting the wrong SQL. Basically it was looking for the QuestionMembeship object by the id column which doesn't exist.
I was under the impression that that column was unnecessary with has_many_through relationships, although it seems I was misguided.
To fix, I simply added the id column to the table as a primary key. Thanks for all the pointers.