Query on freshly inserted Table Entities on Azure storage do not return all entities

Question

Query on freshly inserted Table Entities on Azure storage do not return all entities

Patrick Wellink 20

After inserting a 1000 table entities, the first read gives Less entities. I have a very simple program

        public static async Task RunTest(TableClient client)
        {
            List<TableEntity> tableEntities = new List<TableEntity>();

            // Create 1000 table entities 
            for (int i = 0; i < 1000; i++)
            {
                TableEntity entity = new TableEntity();
                entity.PartitionKey = $"{new Random().Next(1, 10)}";
                entity.RowKey = $"{Guid.NewGuid()}";
                tableEntities.Add(entity);
            }

            Parallel.ForEach(tableEntities, async entity =>
            {
                await client.UpsertEntityAsync(entity);
            });

            for (int i = 0; i < 10; i++)
            {
                List<TableEntity> results = new();
                var asyncdata = client.QueryAsync<TableEntity>(maxPerPage: 200).ConfigureAwait(false);
                await foreach (TableEntity page in asyncdata)
                {
                    results.Add(page);
                }
                Console.WriteLine($"Read entities :{results.Count}");    
            }
        }

I would expect 1000 table entities when reading them. But i get less entities than i inserted. Below is some output from the program.

Read entities :999
Read entities :1000
Read entities :1000
Read entities :1000
Read entities :1000
Read entities :1000
Read entities :1000
Read entities :1000
Read entities :1000
Read entities :1000

The behaviour is pretty consistent, when i run it for a second time:

Read entities :1999
Read entities :2000
Read entities :2000
Read entities :2000
Read entities :2000
Read entities :2000
Read entities :2000
Read entities :2000
Read entities :2000
Read entities :2000

This time there are 1000 new entities added but again not all are read. Sometimes i miss 1 record, or 4, the number differs from time to time.

Sai Raghunadh M 4,575 Reputation points Microsoft External Staff Moderator

2024-10-05T00:20:00.2166667+00:00

Hi @Patrick Wellink,

Thanks for the question and using MS Q&A platform.

A request that returns more than the default maximum or specified maximum number of results returns a continuation token for performing pagination. When making subsequent requests that include continuation tokens, be sure to pass the original URI on the request. For example, if you have specified a $filter, $select, or $top query option as part of the original request, you will want to include that option on subsequent requests. Otherwise your subsequent requests may return unexpected results.

Please go through this document that might help you: Querying tables and entities.

I hope this information helps, please do let us know if you have any Queries.
Patrick Wellink 20 Reputation points

2024-10-05T08:58:14.75+00:00

I do apreciate any help, but your remark told me stuff i already know, the question is, WHY DO I GET 999 Results when there are clearly 1000 records in the table ?
Sai Raghunadh M 4,575 Reputation points Microsoft External Staff Moderator

2024-10-07T19:35:55.7033333+00:00

Hi @Patrick Wellink,

When running the test multiple times, you might notice variations in the output, such as one run returning 999 entities and another returning 1000. This inconsistency suggests that some read operations may be occurring before all the upsert operations have completed and the entities are fully visible.

In cloud services, there can be slight delays in the visibility of newly inserted entities, which may result in fewer entities being read immediately after insertion.

If you frequently encounter fewer entities than expected, consider implementing a retry mechanism or adding a brief delay before the read operations. This can give the system additional time to ensure that all entities have been fully committed and are available for querying.

For more Information on Task: Delay.

I hope this information helps, please do let us know if you have any Queries.
Sai Raghunadh M 4,575 Reputation points Microsoft External Staff Moderator

2024-10-08T21:17:45.0533333+00:00

Hi @Patrick Wellink,

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.
Patrick Wellink 20 Reputation points

2024-10-09T11:37:11.3433333+00:00

No.. I don't have a resolution.......

I am submitting a record to Azure, I await the response. (So all work should be done)

But then Azure has to do some background Syncing (or whatever) , and if i query at that moment, I could not get a result......

This seems very strange to me, and is absolutely not common use in cloud scenarios.....

Suppose Redis would work like that, would be quite worthless...... You can insert, but only read after a little while.......

Or who can guarantee me that the background syncing task would successfully complete in azure. (will the records be visible then ?)
Sai Raghunadh M 4,575 Reputation points Microsoft External Staff Moderator

2024-10-09T17:56:01.3733333+00:00

Hi @Patrick Wellink,

Consistency Models:

Azure services can behave differently when it comes to data consistency. For example, Azure Cosmos DB has various consistency levels like strong, eventual, and others. If you're using a level that allows eventual consistency, you might find that data isn't immediately available after you save it.

Background Operations:

Azure often processes tasks in the background. While services like Azure SQL Database usually show data right after you write it, things like network delays or transaction logs can still cause small delays in visibility.

Retry Logic:

It's a good idea to add a retry mechanism when working with cloud databases. If your first read doesn’t return all the data you expect, waiting a moment and trying again can help.

Testing and Debugging:

Testing under different conditions is important. You should monitor how your application performs and look for patterns that might explain any inconsistencies.

Redis Comparison:

You’re correct that Redis is designed for real-time access and provides strong consistency. This means it generally shows data immediately, making it ideal for low-latency applications.
Patrick Wellink 20 Reputation points

2024-10-09T19:21:52.47+00:00

I know this, but for table service on a storage account I would expect strong consistency
Sai Raghunadh M 4,575 Reputation points Microsoft External Staff Moderator

2024-10-10T00:20:54.0133333+00:00

Hi @Patrick Wellink,

Azure Table Storage is usually consistent, so when you add or update data, you should see it right away. But sometimes, you might run into temporary issues. Here are some things to think about:

Network Delays: If your app and storage are in different regions, there could be delays in seeing the data.

Batch Updates: If you're updating multiple records at once, all updates need to succeed. If one fails, none will show up.

Multiple Users: If many people or processes are reading and writing at the same time, it might cause confusion in what you see.

Error Handling: Make sure to handle errors well. If you see fewer records than expected, log it and try reading again after a short wait.

These issues are rare, but following these tips can help. If problems continue, consider reaching out to Azure support for help.
Sai Raghunadh M 4,575 Reputation points Microsoft External Staff Moderator

2024-10-11T13:34:34.47+00:00

Hi @Patrick Wellink,

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.
Sai Raghunadh M 4,575 Reputation points Microsoft External Staff Moderator

2024-10-14T16:15:14.2933333+00:00

Hi @Patrick Wellink,

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.
Patrick Wellink 20 Reputation points

2024-10-16T14:58:40.9866667+00:00

Well I am just dissapointed in the table Service.

When selling it to a customer they say strong consistency is used : See Advertorial , under Made for the enterprise.

But that is just not true, and only looks nice on paper.
Sai Raghunadh M 4,575 Reputation points Microsoft External Staff Moderator

2024-10-16T18:39:37.9866667+00:00

Hi @Patrick Wellink,

It might be worthwhile to reach out to Azure support to investigate further, especially since you're observing behavior that seems inconsistent with the advertised capabilities of the service.
Sai Raghunadh M 4,575 Reputation points Microsoft External Staff Moderator

2024-10-17T18:44:08.7733333+00:00

Hi @Patrick Wellink,

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.

Answer accepted by question author

0 additional answers

Your answer

Sai Raghunadh M 4,575 Reputation points Microsoft External Staff Moderator

2024-10-05T00:20:00.2166667+00:00

Hi @Patrick Wellink,

Thanks for the question and using MS Q&A platform.

A request that returns more than the default maximum or specified maximum number of results returns a continuation token for performing pagination. When making subsequent requests that include continuation tokens, be sure to pass the original URI on the request. For example, if you have specified a $filter, $select, or $top query option as part of the original request, you will want to include that option on subsequent requests. Otherwise your subsequent requests may return unexpected results.

Please go through this document that might help you: Querying tables and entities.

I hope this information helps, please do let us know if you have any Queries.
Patrick Wellink 20 Reputation points

2024-10-05T08:58:14.75+00:00

I do apreciate any help, but your remark told me stuff i already know, the question is, WHY DO I GET 999 Results when there are clearly 1000 records in the table ?
Sai Raghunadh M 4,575 Reputation points Microsoft External Staff Moderator

2024-10-07T19:35:55.7033333+00:00

Hi @Patrick Wellink,

When running the test multiple times, you might notice variations in the output, such as one run returning 999 entities and another returning 1000. This inconsistency suggests that some read operations may be occurring before all the upsert operations have completed and the entities are fully visible.

In cloud services, there can be slight delays in the visibility of newly inserted entities, which may result in fewer entities being read immediately after insertion.

If you frequently encounter fewer entities than expected, consider implementing a retry mechanism or adding a brief delay before the read operations. This can give the system additional time to ensure that all entities have been fully committed and are available for querying.

For more Information on Task: Delay.

I hope this information helps, please do let us know if you have any Queries.
Sai Raghunadh M 4,575 Reputation points Microsoft External Staff Moderator

2024-10-08T21:17:45.0533333+00:00

Hi @Patrick Wellink,

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.
Patrick Wellink 20 Reputation points

2024-10-09T11:37:11.3433333+00:00

No.. I don't have a resolution.......

I am submitting a record to Azure, I await the response. (So all work should be done)

But then Azure has to do some background Syncing (or whatever) , and if i query at that moment, I could not get a result......

This seems very strange to me, and is absolutely not common use in cloud scenarios.....

Suppose Redis would work like that, would be quite worthless...... You can insert, but only read after a little while.......

Or who can guarantee me that the background syncing task would successfully complete in azure. (will the records be visible then ?)
Sai Raghunadh M 4,575 Reputation points Microsoft External Staff Moderator

2024-10-09T17:56:01.3733333+00:00

Hi @Patrick Wellink,

Consistency Models:

Azure services can behave differently when it comes to data consistency. For example, Azure Cosmos DB has various consistency levels like strong, eventual, and others. If you're using a level that allows eventual consistency, you might find that data isn't immediately available after you save it.

Background Operations:

Azure often processes tasks in the background. While services like Azure SQL Database usually show data right after you write it, things like network delays or transaction logs can still cause small delays in visibility.

Retry Logic:

It's a good idea to add a retry mechanism when working with cloud databases. If your first read doesn’t return all the data you expect, waiting a moment and trying again can help.

Testing and Debugging:

Testing under different conditions is important. You should monitor how your application performs and look for patterns that might explain any inconsistencies.

Redis Comparison:

You’re correct that Redis is designed for real-time access and provides strong consistency. This means it generally shows data immediately, making it ideal for low-latency applications.
Patrick Wellink 20 Reputation points

2024-10-09T19:21:52.47+00:00

I know this, but for table service on a storage account I would expect strong consistency
Sai Raghunadh M 4,575 Reputation points Microsoft External Staff Moderator

2024-10-10T00:20:54.0133333+00:00

Hi @Patrick Wellink,

Azure Table Storage is usually consistent, so when you add or update data, you should see it right away. But sometimes, you might run into temporary issues. Here are some things to think about:

Network Delays: If your app and storage are in different regions, there could be delays in seeing the data.

Batch Updates: If you're updating multiple records at once, all updates need to succeed. If one fails, none will show up.

Multiple Users: If many people or processes are reading and writing at the same time, it might cause confusion in what you see.

Error Handling: Make sure to handle errors well. If you see fewer records than expected, log it and try reading again after a short wait.

These issues are rare, but following these tips can help. If problems continue, consider reaching out to Azure support for help.
Sai Raghunadh M 4,575 Reputation points Microsoft External Staff Moderator

2024-10-11T13:34:34.47+00:00

Hi @Patrick Wellink,

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.
Sai Raghunadh M 4,575 Reputation points Microsoft External Staff Moderator

2024-10-14T16:15:14.2933333+00:00

Hi @Patrick Wellink,

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.
Patrick Wellink 20 Reputation points

2024-10-16T14:58:40.9866667+00:00

Well I am just dissapointed in the table Service.

When selling it to a customer they say strong consistency is used : See Advertorial , under Made for the enterprise.

But that is just not true, and only looks nice on paper.
Sai Raghunadh M 4,575 Reputation points Microsoft External Staff Moderator

2024-10-16T18:39:37.9866667+00:00

Hi @Patrick Wellink,

It might be worthwhile to reach out to Azure support to investigate further, especially since you're observing behavior that seems inconsistent with the advertised capabilities of the service.
Sai Raghunadh M 4,575 Reputation points Microsoft External Staff Moderator

2024-10-17T18:44:08.7733333+00:00

Hi @Patrick Wellink,

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.

Answer 1

TP 163.5K Volunteer Moderator

Hi Patrick,

You are seeing incorrect results because you are querying the table before your code has finished inserting all of the entities. Please test using modified code below:

public static async Task RunTest(TableClient client)
{
    List<TableEntity> tableEntities = new List<TableEntity>();

    // Create 1000 table entities 
    for (int i = 0; i < 1000; i++)
    {
        TableEntity entity = new TableEntity();
        entity.PartitionKey = $"{new Random().Next(1, 10)}";
        entity.RowKey = $"{Guid.NewGuid()}";
        tableEntities.Add(entity);
    }

    await Parallel.ForEachAsync(tableEntities, async (entity, token) =>
        {
            await client.UpsertEntityAsync(entity);
        });

    for (int i = 0; i < 10; i++)
    {
        List<TableEntity> results = new();
        var asyncdata = client.QueryAsync<TableEntity>(maxPerPage: 200).ConfigureAwait(false);
        await foreach (TableEntity page in asyncdata)
        {
            results.Add(page);
        }
        Console.WriteLine($"Read entities :{results.Count}");
    }
}

Please click Accept Answer and upvote if the above was helpful.

Thanks.

-TP

Sai Raghunadh M 4,575 Reputation points Microsoft External Staff Moderator

2024-10-21T17:48:02.32+00:00

Hi @Patrick Wellink,

Following up to see if the above answer provided by @TP was helpful. If this answers your query, do click Accept Answer and Yes for was this answer helpful. And, if you have any further query do let us know.
TP 163.5K Reputation points Volunteer Moderator

2024-10-22T22:34:14.5766667+00:00

@Patrick Wellink Any update? Did you get a chance to run the code I posted above?
Patrick Wellink 20 Reputation points

2024-10-23T18:36:15.5833333+00:00

Had no time yet t o look at it, but will do...