my PostgreSQL 9.2 database for some reason skips an ID with every record. Example:
User
1258930
1258932
1258934
1258936
What would cause this? Any pointer in the right direction to resolve this issue is appreciate. Thank you
The comments have adequately covered possible reasons:
multiple nextval calls, say one by a default and one explicit;
The same sequence used by more than one table
Transactions being rolled back
Any pointer in the right direction to resolve this issue is appreciated
Your key mistake is viewing this as an issue. It's entirely normal for gaps to appear in generated sequences. If your DB crashes and restarts, a gap will appear in a sequence. If a transaction rolls back after allocating IDs, a gap will appear in the sequence.
Your application must be able to deal with this. It shouldn't care what an ID is, only that it's unique.
For details and for hints if you need truly gapless sequences, see this answer.
Add to Craig's points the key culprit I think you will find is a DO ALSO rule inserting the new.id into another table. This will cause double incrementing as RULES rewrite queries so new.id ends up being "whatever we did to calculate new.id last time" (which means incrementing the sequence again).
if it is consistent, the most likely cause IMO is someone creating DO ALSO rules without fully understanding the gotchas.
Related
In InfluxDB (1.5), I have a table where the fields have become inconsistently typed. Most rows in the table are Integer, however, some rows have become strings.
How is this possible? I thought, once a field's types were set (upon first insert), any insert into the table with incorrect typing would fail.
What do I do now? If I go back and attempt to overwrite the data in the inconsistent rows, I get errors saying the field is a string.
After some more research, here's what I've discovered:
Answer to Part 1:
InfluxDB uses a system they refer to as 'sharding' - while I don't know the specifics, I do know that data from the same measurement/table can be stored across multiple, different 'shards'.
According to the InfluxDB documentation, field types can differ between these shards, within the same field, on the same table.
Answer to Part 2:
In order to fix this, the currently-suggested answer is to make a new table, download all the data, and re-insert while ensuring the data that gets inserted is the proper types.
If you had a tag which changed type and became a field, this can be especially difficult to fix, the link above does not address that. To do selects only on tag or field, you can use tag_name::tag or field_name::field within a select statement.
The GROUP BY * clause suggested in the link is required in order to preserve tags, but seemed to cause issues when I used it.
My current solution is a PHP script that uses curl, downloads the points, chunks them, then re-inserts the points into the new table, ensuring each point that gets inserted is casted to the new, uniform type, and properly inserted.
The best way to stopping future issues, is simply not to have them. I went looking for how to lock field types in all cases, across all shards, for a particular measurement table.
Unfortunately, it seems impossible to guarantee 100% type consistency across all current and future shards. "Don't make mistakes because it's really difficult to clean up" seems to be InfluxDB's modus operandi.
First of all I request you to please bear with me and I apologise if this is a silly question.
I have a table like this.
create table users (
cname text,
--anything else like counter or timestamp
primary key (cname)
);
All I need to do is implement a stack like structure with that table.
A number of insert and delete operations will be there.
Problems faced:
1) I tried using timestamps. I successfully inserted using dateof(now()) but I just wanted to delete the last 1 record (???). Also, by making cname,t (where t is of type timestamp) as primary key, I'm having redundant cnames which I don't want.
2) I tried using counter but I felt it was complicated. Also, I may have multiple threads or clients performing the insert/delete operations. So, I thought leave it.
3) Also, I will not be knowing the value of cname. So, queries which require key in where clause are impossible. So i think I need to change the primary key to some other variable.
Please help me move forward. I'm finding it a bit difficult as there are no good books available for cql to learn from.
Stacks and Queues are antipatterns in cassandra ( http://www.datastax.com/dev/blog/cassandra-anti-patterns-queues-and-queue-like-datasets ) - you can implement them, but you need to understand the internals quite well before you'll be able to do it without making a horrible mistake. It's fairly clear that you're probably not at that level yet.
Is there a reason you think you need to use cassandra for a stack?
Yes, the first part of your primary key (known as your partition key) needs to be something you know, so if you dont know cname, it's probably not a great fit for your primary key. What DO you know about the data/usage patterns? Maybe time buckets?
MongoDB uses ObjectId type for _id.
Will it be bad if I make _id an incrementing integer?
(With this gem, if you're interested)
No it isn't bad at all and in fact the built in ObjectId is quite sizeable within the index so if you believe you have something better then you are more than welcome to change the default value of the _id field to whatever.
But, and this is a big but, there are some considerations when deciding to move away from the default formulated ObjectId, especially when using the auto incrementing _ids as shown here: https://docs.mongodb.com/v3.0/tutorial/create-an-auto-incrementing-field
Multi threading isn't such a big problem because findAndModify and the atomic locks can actually take care of that, but then you just hit into your first problem. findAndModify is not the fastest function nor the lightest and there have been significant performance drops noticed when using it regularly.
You also have to consider the overhead of doing this yourself anyway, even without findAndModify. For every insert you will need an extra query. Imagine having a unique id that you have to query the uniqueness of every time you want to insert. Eventually your insert rate will drop to a crawl and your lock time will build up.
Of course the ObjectId is really good at being unique without having to check or formulate its own uniqueness by touching the database prior to insertion, hence it doesn't have this overhead.
If you still feel an integer _id suites your scenario, then go for it, but bare in mind the overhead described above.
You can do it, but you are responsible to make sure that the integers are unique.
MongoDB doesn't support auto-increment fields like most SQL databases. When you have a distributed or multithreaded application which has multiple processes and/or threads which create new database entries, you have to make sure that they use the same counter. Otherwise it could happen that two threads try to store a document with the same _id in the database.
When that happens, one of them will fail. That means you have to wait for the database to return a success or error (by calling GetLastError or by setting the write concerns to acknowledged), which takes longer than just sending data in a fire-and-forget manner.
I had a use case for this: replacing _id with a 64 bit integer that represented a simhash of a document index for searching.
Since I intended to "Get or create", providing the initial simhash, and creating a new record if one didn't exist was perfect. Also, for anyone Googling, MongoDB support explained to me that simhashes are absolutely perfect for sharding and scaling, and even better than the more generic ObjectId, because they will divide up the data across shards perfectly and intrinsically, and you get the key stored for negative space (a uint64 is much smaller than an objectId and would need to be stored anyway).
Also, for you Googlers, replacing a MongoDB _id with something other than an objectId is absolutely simple: Just create an object with the _id being defined; use an integer if you like. That's it: Mongo will simply use it. If you try to create a document with the same _id you'll get an error (E11000/Duplicate key). So like me, if you're using simhashing, this is ideal in all respects.
The Scenario
Update: It was brought to my attention that ordering by created_at will actually compare a millisecond float that's of sufficient resolution (by far). However, while I feel a bit dumb now, my question still stands. My scenario is just irrelevant, so I removed it.
The Question
I know that the database knows precisely the order of creation by tracking a row's ID.
Are there any pitfalls in relying on latest ID to determine order?
A better solution is to replace the latest_post_at with something more precise than a second. Time.now.to_f instead of .to_i will give you sub-second precision (millisecond I think, the docs aren't clear). Should two posts happen to have the same millisecond timestamp you could use the id as a tie-breaker.
If you're using whatever is the "natural" way of generating autoincrementing surrogate primary keys for your database, the only pitfall that comes to mind is that the order in which the database sequencer generated the IDs might not be the order in which the transactions that create the Post records start or finish. (Or however you define the time when a post is "created".)
Considering the transaction should normally take a fraction of a second to complete this uncertainty might be irrelevant for your needs.
In my present Rails application, I am resolving scheduling conflicts by sorting the models by the "created_at" field. However, I realized that when inserting multiple models from a form that allows this, all of the created_at times are exactly the same!
This is more a question of best programming practices: Can your application rely on your ID column in your database to increment greater and greater with each INSERT to get their order of creation? To put it another way, can I sort a group of rows I pull out of my database by their ID column and be assured this is an accurate sort based on creation order? And is this a good practice in my application?
The generated identification numbers will be unique.
Regardless of whether you use Sequences, like in PostgreSQL and Oracle or if you use another mechanism like auto-increment of MySQL.
However, Sequences are most often acquired in bulks of, for example 20 numbers.
So with PostgreSQL you can not determine which field was inserted first. There might even be gaps in the id's of inserted records.
Therefore you shouldn't use a generated id field for a task like that in order to not rely on database implementation details.
Generating a created or updated field during command execution is much better for sorting by creation-, or update-time later on.
For example:
INSERT INTO A (data, created) VALUES (smething, DATE())
UPDATE A SET data=something, updated=DATE()
That depends on your database vendor.
MySQL I believe absolutely orders auto increment keys. SQL Server I don't know for sure that it does or not but I believe that it does.
Where you'll run into problems is with databases that don't support this functionality, most notably Oracle that uses sequences, which are roughly but not absolutely ordered.
An alternative might be to go for created time and then ID.
I believe the answer to your question is yes...if I read between the lines, I think you are concerned that the system may re-use ID's numbers that are 'missing' in the sequence, and therefore if you had used 1,2,3,5,6,7 as ID numbers, in all the implementations I know of, the next ID number will always be 8 (or possibly higher), but I don't know of any DB that would try and figure out that record Id #4 is missing, so attempt to re-use that ID number.
Though I am most familiar with SQL Server, I don't know why any vendor who try and fill the gaps in a sequence - think of the overhead of keeping that list of unused ID's, as opposed to just always keeping track of the last I number used, and adding 1.
I'd say you could safely rely on the next ID assigned number always being higher than the last - not just unique.
Yes the id will be unique and no, you can not and should not rely on it for sorting - it is there to guarantee row uniqueness only. The best approach is, as emktas indicated, to use a separate "updated" or "created" field for just this information.
For setting the creation time, you can just use a default value like this
CREATE TABLE foo (
id INTEGER UNSIGNED AUTO_INCREMENT NOT NULL;
created TIMESTAMP NOT NULL DEFAULT NOW();
updated TIMESTAMP;
PRIMARY KEY(id);
) engine=InnoDB; ## whatever :P
Now, that takes care of creation time. with update time I would suggest an AFTER UPDATE trigger like this one (of course you can do it in a separate query, but the trigger, in my opinion, is a better solution - more transparent):
DELIMITER $$
CREATE TRIGGER foo_a_upd AFTER UPDATE ON foo
FOR EACH ROW BEGIN
SET NEW.updated = NOW();
END;
$$
DELIMITER ;
And that should do it.
EDIT:
Woe is me. Foolishly I've not specified, that this is for mysql, there might be some differences in the function names (namely, 'NOW') and other subtle itty-bitty.
One caveat to EJB's answer:
SQL does not give any guarantee of ordering if you don't specify an order by column. E.g. if you delete some early rows, then insert 'em, the new ones may end up living in the same place in the db the old ones did (albeit with new IDs), and that's what it may use as its default sort.
FWIW, I typically use order by ID as an effective version of order by created_at. It's cheaper in that it doesn't require adding an index to a datetime field (which is bigger and therefore slower than a simple integer primary key index), guaranteed to be different, and I don't really care if a few rows that were added at about the same time sort in some slightly different order.
This is probably DB engine depended. I would check how your DB implements sequences and if there are no documented problems then I would decide to rely on ID.
E.g. Postgresql sequence is OK unless you play with the sequence cache parameters.
There is a possibility that other programmer will manually create or copy records from different DB with wrong ID column. However I would simplify the problem. Do not bother with low probability cases where someone will manually destroy data integrity. You cannot protect against everything.
My advice is to rely on sequence generated IDs and move your project forward.
In theory yes the highest id number is the last created. Remember though that databases do have the ability to temporaily turn off the insert of the autogenerated value , insert some records manaully and then turn it back on. These inserts are no typically used on a production system but can happen occasionally when moving a large chunk of data from another system.