I've cloned a repo from here and trying to explore AWS AppSync's subscription. My understanding is that if there is real-time updates to server data, client should expect to see some notification or updates or sorts, so what I did was:
running the app on a simulator
Open DynamoDB console and add the records manually.
I was expecting there is some notifications received on my app but there isn't, and if I refresh the app it will have the updated records? Am I understand the subscription wrongly?
Subscriptions are not triggered from your dynamo db, but from your mutations (defined in your graphql schema). Try to add records via the mutation your subscription listens on. You can run a mutation from the app sync console under "queries".
If your client is set up correctly, it should update accordingly.
Hope this helps :)
Subscription can only be triggered by mutation. When you add record directly to your DB, the mutation is not called hence no subscription is triggered. Does not really servers the purpose for the external db updates. There is work around available.
Scenario 1: If you are making the change to the DB directly via some DB client, you need to call the mutation endpoint explicitly (from AWS console, postman etc.). This will trigger the subscription. I am guessing the direct DB change is done for testing.
Scenario 2: The direct DB change is done by some other external process and not via Appsync mutation. You need to call the none data source mapped mutation in your processor. This dummy mutation will trigger the subscription.
Here's [a link] (https://aws.amazon.com/premiumsupport/knowledge-center/appsync-notify-subscribers-real-time/ ) explaining how to create a none data source mapped mutation.
Related
I deleted all the items in the DataTemplate table but when I query them again with the searchDataTemplates endpoint on the app or in AppSync it returns the old data, but when I use the listDataTemplates it returns nothing which is correct. Needed to repopulate the data in the table.
data template table
search endpoint
list endpoint
when I updated items individually it worked just fine but when i deleted all the items from the console (around 700 items) the search endpoint stopped working. Just the search
UPDATE:
I repopulated the data hoping it'd reset but now the listDataTemplates shows the new data and the search still shows the old data, is there some cache that needs to be reset?
SECOND UPDATE:
I removed the table and the appsync functions are gone however when i recreated the table (with no data) the testing out the function still returns the old data. I'm guessing the opensearch stuff hasn't been updated?
If you are using AppSync with Amplify CLI, #searchable will automatically create the followings:
An OpenSearch Domain
A Lambda Function that will be attached to the DynamoDB Streams and push the changes (create/update/delete) over to your OpenSearch Domain.
And the problem that you're facing is most likely due to the Lambda Function created failed to push the changes from DynamoDB Streams to OpenSearch. A quick suggestion is to check on the created Lambda Function first.
Reference: #searchable
This issue can only happen if caching is enabled in your application.
I am not sure what's the infrastructure you are using, so i would go ahead with some educated guess. Please feel free to correct me if i overstepped.
From your description of question, you have an AppSync as API layer and DynamoDb as primary database.
If these are the only two resources you have, please check the AppSync cache configuration.
Open AppSync console
from left panel select APIs -> your api -> caching
Validate Caching behavior is set to None
In case if you have AWS OpenSearch enabled for search query (i could be wrong, however picking up from previous comment). Then validate the cluster configuration.
Open AWS Open Search Service console
From left panel select Domains and click on the openserch domain that you are using
scroll to the bottom right and look for Advanced cluster settings and ensure the attribute Fielddata cache allocation is set to 0
If Fielddata cache allocation is not 0, update the cluster configuration and modify the advanced cluster setting to set the Fielddata cache allocation field to 0.
Wait for a few minutes (I would suggest 5 minutes) and then retry your use-case.
I hope this would help resolve your issue.
I am working on the requirement where the data entered in the form needs to be validated manually and once validated , a approval mail be be sent out and then data will be stored in the database.I plan to use AWS step function for this with token.
https://aws.amazon.com/blogs/compute/implementing-serverless-manual-approval-steps-in-aws-step-functions-and-amazon-api-gateway/
I plan to use a similar design like in the link above.However is there a way not to use API Gateway for sending back the task token to step function to resume processing.Did anybody worked on the similar requirement and how the functionality was achieved. Thank you.
Step function can be invoked by the AWS Lambda function as well.
Once the form is validated and stored in database, you can trigger the Lambda function based on the database events(ex- if DynamoDB used then based on the DynamDB streams), and the lambda can start the step function.
I have a Process Instance that was started by the Tenant 949.
I tried to fetch and lock that Task, like described here: https://docs.camunda.org/manual/7.10/reference/rest/external-task/fetch/
Here is the Body of the Request:
{"workerId":"testUser","maxTasks":1,"usePriority":false,
"topics":[
{"topicName":"archive-document","tenantIdIn":["949"],"lockDuration":10000,"localVariables":true,"deserializeValues":false}
]}
I don't get any Task with it.
The same request works if the Process Instance is started without a Tenant and fetched accordingly.
Do I miss something, or is this a Bug of Camunda?
Have you attempted to simply do a query to first retrieve the task? (Rather than attempting to fetch it and lock it?) You could use this endpoint: https://docs.camunda.org/manual/7.10/reference/rest/external-task/get-query/.
You may also want to query the runtime database directly using SQL. Your External Task would be in the ACT_RU_EXT_TASK table and would have a TOPIC_NAME_ defined within it (as well as a TENANT_ID_).
The problem was the Authentication.
I had a different User to start the process and to fetch the Task.
And this User had no rights to fetch the Task for this Tenant.
I don't know how to subscribe to changes in DynamoDB database. Let me show an example: User A sends a message (which is saved in the database) to User B and in the User B's app the message automatically appears.
I know this is possible with recently released AWS AppSync, but I couldn't integrate it with Ionic (which I am using). However, there must be an alternative since AWS AppSync was released only at the end of 2017/beginning of 2018.
I've also seen something called Streams in DynamoDB but not sure if that's what I need.
DynamoDB Streams is designed specifically for capturing/subscribing to table activity. You can set up a Lambda Function with your notification logic to process the stream and send notifications accordingly.
Suppose I have a task of updating an user via a third party API call. Is it okay to put the actual user data inside the message (if it fits)? Or should I only provide an ID in the message so the worker can retrieve the updated record from my local database?
You need to check what level of compliance is required for your infrastructure, to see what kind of data you want to put in the queue.
If there aren't any compliance restrictions, you are free to put any kind of data in your own infrastructure on AWS.