I have the following setup where a service layer, using jooq, contacts a PostgreSQL database.
In this scenario, whenever multiple requests happen quickly one after another (or even not that quickly), I get the following error message:
Internal error processing createItem: Cannot get autoCommit
My queries all run within transactions (using jooq's transactionResult methods).
Searching has not yielded many results, and I do not see why autoCommit should even be enabled in those cases. Is this most likely a configuration issue, or is there something else I can try to troubleshoot this issue better?
I noticed the same problem and message when running massive batch uploads on the limit of physical memory and limited amount of db connection (specific to my environment). It would be hard to provide a reproduction case for that, but to me this is a sign of db performance/memory starvation. Reduction of Java execution threads helped in my case.
Related
I am architecting a database where I expected to have 1,000s of tenants where some data will be shared between tenants. I am currently planning on using Postgres with row level security for tenant isolation. I am also using knex and Objection.js to model the database in node.js.
Most of the tutorials I have seen look like this where you create a separate knex connection per tenant. However, I've run into a problem on my development machine where after I create ~100 connections, I received this error: "remaining connection slots are reserved for non-replication superuser connections".
I'm investigating a few possible solutions/work-arounds, but I was wondering if anyone has been able to make this setup work the way I'm intending. Thanks!
Perhaps one solution might be to cache a limited number of connections, and destroy the oldest cached connection when the limit is reached. See this code as an example.
That code should probably be improved, however, to use a Map as the knexCache instead of an object, since a Map remembers the insertion order.
I have taken over an existing MVC website which uses entity framework and hangfire and is hosted on Azure and uses Azure database. Every so often the website times out.
I'm new to Azure portal, entity framework and hangfire.
If I increase the DTU's it clears the timeout issues?
I'm looking for ways of how to diagnose why the website times out. I have added error logging using elmah and checked hangfire but this doesn't give me any further information.
Is there anything in azure portal that can help?
If it "times out" and if "increasing DTU resolves timeouts" and these observations are true (I think it's on you to really convince yourself this is absolutely true, don't make this assumption lightly) then the usual and obvious candidate is "a slow sql query". Entity Framework is often used with linq to create sql queries without writing sql. These queries are often fine for very simple tasks, such as someData.Where(x=>x.Id == 1).First(), however, if linq is used to join tables, or create complex associations, the generated sql can become monstrously bad, from a performance perspective. You can add logging to write out the sql generated by linq, or you can try to trace the database to see what sql is running on it. If tracing is out of the question, there are still meta queries you can use to view things like cached query plans and SQL Server can give you estimated costs and cached execution counts.
You can still hang yourself without using linq. You can still use stored procedures with EF. Way too many developers are naive about SQL performance still; you need to comb over your back end and learn the schema, the stored procedures; inspect the sql contents of everything. Check for any database triggers (easy to miss). Red flags are subqueries, too many joining, too many results from a query, lots of string manipulation in a query, joining tables on strings, or XML/JSON-based SQL work.
Be aware that "slow sql queries" will become slower when load is high. And when slow sql queries build up, they only take more time to resolve. This can also cause debilitating table locking, depending on the nature of the query.
But queries can be performant and still cause locking. ie One table is being written to often and it's blocking other writes or reads from that table. This is a little harder to diagnose, but you can figure it out by carefully inspecting logs of database calls and how long they take to execute. There are also sql queries you can run on the database to diagnose long-running queries, or what tables are locked at a given point in time.
Finally, check for any back end webjobs for your application. If timeouts occur at reoccurring days or times, then somebody's batch SQL could be blocking your production database from being read.
But this is all speculation. I think you need to do more research to determine what is actually causing the site to become unresponsive. If you can log response times for common queries, you can rule out SQL-based latency as being the culprit or not and work from there. There's nothing inherently "amiss" about any of the technologies you specified.
If queries are perfomant but still causing issues, a long term solution is to add something like a message queue and batch your sql work intelligently, or just make the database work asynchronous and not block the UI.
You should correlate any logged timeouts with azure's monitoring. Azure can give you CPU/RAM/page visits and such on the dashboard.
SQL Azure is a bit of a different beast. It doesn't have the on-demand performance of a dedicated DB unless you're prepared to throw serious $$ at it. And even then ...
EF, when written for well can perform quite well. When written poorly it can be a dog, and those problems are compounded on a platform like SQL Azure.
The first thing is to check that your EF contexts are set up to use an execution strategy suited to Azure: https://learn.microsoft.com/en-us/ef/ef6/fundamentals/connection-resiliency/retry-logic
The next thing would be to see what kinds of SQL tracing you can run on Azure. Tracing is essential to see what EF is doing behind the scenes. I'm not familiar with tools available for Azure, in my case my Azure experience was running SQL Server on VMs because SQL Azure was too immature, not HIPAA compliant at the time, and expensive for the DTU estimates we were able to get. Worst case, can you restore an database backup into an SQL Server instance and point a copy of your application environment temporarily at that to run through common usage scenarios? Using an SQL Trace you can pick up on exactly when and how often EF is executing queries, and what queries it is executing.
Things to look at:
How many queries are running? If you are loading a set of records and expect one query, are there a whole heap of queries getting sent? This would indicate lazy-load calls being triggered.
What queries are being run? Is it selecting a lot more fields than are being displayed? This would be potentially a case where entire entities are being loaded where a .Select() could be used to reduce the amount of data. Perhaps even the case where entire sets of entities are being loaded that aren't relevant to what is displayed/done, such as cases where someone is using .ToList() prior to just doing a .Count() or .Any() or doing a .FirstOrDefault() just to do a != null check.
Is the database properly indexed? Copy some of the heavier queries into SQL Manager and execute them with an execution plan. Are there indexing suggestions?
The common sins of developing with EF and other ORMs boil down to "pulling too much, too often." It's surprising how many clients I've worked with have development teams that have not used a profiler to inspect their ORM use efficiency. (and I'm talking 0% so far.)
I'm creating a reporting engine that makes a couple of long queries over a standby server and process the result with pandas. Everything works fine but sometimes I have some issues with the execution of those queries using a psycopg2 cursor: the query is cancelled with the following message:
ERROR: cancelling statement due to conflict with recovery
Detail: User query might have needed to see row versions that must be removed
I was investigating this issue
PostgreSQL ERROR: canceling statement due to conflict with recovery
https://www.postgresql.org/docs/9.0/static/hot-standby.html#HOT-STANDBY-CONFLICT
but all solutions suggest fixing the issue making modifications to the server's configuration. I can't make those modifications (We won the last football game against IT guys :) ) so I want to know how can I deal with this situation from the perspective of a developer. Can I resolve this issue using python code? My temporary solution is simple: catch the exception and retry all the failed queries. Maybe could be done better (I hope so).
Thanks in advance
There is nothing you can do to avoid that error without changing the PostgreSQL configuration (from PostgreSQL 9.1 on, you could e.g. set hot_standby_feedback to on).
You are dealing with the error in the correct fashion – simply retry the failed transaction.
The table data on the hot standby slave server is modified while a long running query is running. A solution (PostgreSQL 9.1+) to make sure the table data is not modified is to suspend the replication on the slave and resume after the query.
select pg_xlog_replay_pause(); -- suspend
select * from foo; -- your query
select pg_xlog_replay_resume(); --resume
I recently encountered a similar error and was also in the position of not being a dba/devops person with access to the underlying database settings.
My solution was to reduce the time of the query where ever possible. Obviously this requires deep knowledge of your tables and data, but I was able to solve my problem with a combination of a more efficient WHERE filter, a GROUPBY aggregation, and more extensive use of indexes.
By reducing the amount of server side execute time and data, you reduce the chance of a rollback error occurring.
However, a rollback can still occur during your shortened window, so a comprehensive solution would also make use of some retry logic for when a rollback error occurs.
Update: A colleague implemented said retry logic as well as batching the query to make the data volumes smaller. These three solutions have made the problem go away entirely.
I got the same error. What you CAN do (if the query is simple enough), is deviding the data into smaller chunks as a workaround.
I did this within a python loop to call the query multiple times with the LIMIT and OFFSET parameter like:
query_chunk = f"""
SELECT *
FROM {database}.{datatable}
LIMIT {chunk_size} OFFSET {i_chunk * chunk_size}
"""
where database and datatable are the names of your sources..
The chunk_size is individually and to set this to a not too high value is crucial for the query to finish.
We decided to use a micro-orm against an Azure Database. As our business only needs "inserts" and "selects", we decided to suppress all code-managed SqlTransaction (no concurrency issues on data).
Then, we noticed that our instance of Azure Database responded very slowly. The "rpc completed" event occured in delays that are hundreds times the time needed to run a simple sql statement.
Next, we benchmarked our code with EF6 and we saw that the server responded very quickly. As EF6 implements a built-in transaction, we decided to restore the SqlTransaction (ReadCommited) on the micro-orm and we noticed everything was fine.
Does Azure Database require an explicit SqlTransaction (managed by code) ? How does the SqlTransaction influence Azure Database performances ? Why was it implemented that way ?
EDIT : I am going to post some more precise information about the way we collected traces. It seems our Azure events logs sometimes express in nanoseconds, sometimes in milliseconds. Seems so weird.
If I understand what you are asking correctly, batching multiple SQL queries into one transaction will give you better results on any DBS. Committing after every insert/update/delete has a huge overhead on a DBS that is not designed for it (like MyISAM on MySQL).
It can even cause bad flushes to disk and thrashing if you do too much. I once had a programmer committing thousands of entries to one of my DBs every minute, each as their own transactions, and it brought the server to a halt.
InnoDB, one of 2 most popular database formats for MySQL, can only commit 20-30 transactions a second (or maybe it was 2-3... it's been a long time), as each is flushed to the disk at the end for ACID compliance.
i know this has been asked here. But my question is slightly different. When the dataset was designed keeping the disconnected principle in mind, what was provided as a feature which would handle unexpected termination of the application, say a power failure or a windows hang or system exception leading to restart. Say the user has entered some 100 rows and it is modified at the dataset alone. Usually the dataset is updated at the application close or at a timely period.
In old times which programming using vb 6.0 all interaction used to take place directly with the database, thus each successful transaction was committing itself automatically. How can that be done using datasets?
DataSets are never for direct access to database, they are a disconnected model only. There is no intent that they be able to recover from machine failures.
If you want to work live against the database you need to use DataReaders and issue DbCommands against the database live for changes. This of course will increase your load on the database server though.
You have to balance the two for most applications. If you know a user just entered vital data as a new row, execute an insert command to the database, and put a copy in your local cached DataSet. Then your local queries can run against the disconnected data, and inserts are stored immediately.
A DataSet can be serialized very easily, so you could implement your own regular backup to disk by using serialization of the DataSet to the filesystem. This will give you some protection, but you will have to write your own code to check for any data that your application may have saved to disk previously and so on...
You could also ignore DataSets and use SqlDataReaders and SqlCommands for the same sort of 'direct access to the database' you are describing.