Re: tombstones problem with 1.0.8
(Radim: I'm assuming you mean do not delete already deleted columns as Ross doesn't delete his rows.) Just to be clear about Ross' situation: he continually inserts columns and later deletes columns from the same set of rows. As long as he *doesn't* *keep deleting already-deleted columns* (which refreshes the tombstone on them), the deleted columns *should* get cleaned up, right? (Even though the row itself continually gets new columns inserted and other columns deleted?) Thanks, John On Tue, Mar 27, 2012 at 2:21 AM, Radim Kolar h...@filez.com wrote: Dne 27.3.2012 11:13, Ross Black napsal(a): Any pointers on what I should be looking for in our application that would be stopping the deletion of tombstones? do not delete already deleted rows. On read cassandra returns deleted rows as empty in range slices.
Internal error processing get_slice (NullPointerException)
Has anyone seen this particular NPE before from Cassandra? This is on 1.0.8. It seems to happen transiently on multiple nodes in my cluster, every so often, and goes away. ERROR [Thrift:45] 2012-03-26 19:59:12,024 Cassandra.java (line 3041) Internal error processing get_slice java.lang.NullPointerException at org.apache.cassandra.db.SliceFromReadCommand.maybeGenerateRetryCommand(SliceFromReadCommand.java:76) at org.apache.cassandra.service.StorageProxy.fetchRows(StorageProxy.java:724) at org.apache.cassandra.service.StorageProxy.read(StorageProxy.java:564) at org.apache.cassandra.thrift.CassandraServer.readColumnFamily(CassandraServer.java:128) at org.apache.cassandra.thrift.CassandraServer.getSlice(CassandraServer.java:283) at org.apache.cassandra.thrift.CassandraServer.multigetSliceInternal(CassandraServer.java:365) at org.apache.cassandra.thrift.CassandraServer.get_slice(CassandraServer.java:326) at org.apache.cassandra.thrift.Cassandra$Processor$get_slice.process(Cassandra.java:3033) at org.apache.cassandra.thrift.Cassandra$Processor.process(Cassandra.java:2889) at org.apache.cassandra.thrift.CustomTThreadPoolServer$WorkerProcess.run(CustomTThreadPoolServer.java:187) at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:886) at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:908) at java.lang.Thread.run(Thread.java:662) The line in question is (I think) the one below, so it looks like the column family reference for a row can sometimes be null? int liveColumnsInRow = row != null ? row.cf.getLiveColumnCount() : 0; Thanks, John
Re: Composite keys and range queries
Hmm, now I'm really confused. This may be of use to you http://www.datastax.com/dev/blog/schema-in-cassandra-1-1 This article is what I actually used to come up with my schema here. In the Clustering, composite keys, and more section they're using a schema very similarly to how I'm trying to use it. They define a composite key with two parts, expecting the first part to be used as the partition key and the second part to be used for ordering. The hash for (uuid-1 , p1) may be 100 and the hash for (uuid-1, p2) may be 1 . Why? Shouldn't only uuid-1 be used as the partition key? (So shouldn't those two hash to the same location?) I'm thinking of using supercolumns for this instead as I know they'll work (where the row key is the uuid and the supercolumn name is the priority), but aren't composite row keys supposed to essentially replace the need for supercolumns? Thanks, and sorry if I'm getting this all wrong, John On Wed, Mar 14, 2012 at 12:52 AM, aaron morton aa...@thelastpickle.comwrote: You are seeing this http://wiki.apache.org/cassandra/FAQ#range_rp The hash for (uuid-1 , p1) may be 100 and the hash for (uuid-1, p2) may be 1 . You cannot do what you want to. Even if you passed a start of (uuid1,empty) and no finish, you would not only get rows where the key starts with uuid1. This may be of use to you http://www.datastax.com/dev/blog/schema-in-cassandra-1-1 Or you can store all the priorities that are valid for an ID in another row. Cheers - Aaron Morton Freelance Developer @aaronmorton http://www.thelastpickle.com On 14/03/2012, at 1:05 PM, John Laban wrote: Forwarding to the Cassandra mailing list as well, in case this is more of an issue on how I'm using Cassandra. Am I correct to assume that I can use range queries on composite row keys, even when using a RandomPartitioner, if I make sure that the first part of the composite key is fixed? Any help would be appreciated, John On Tue, Mar 13, 2012 at 12:15 PM, John Laban j...@pagerduty.com wrote: Hi, I have a column family that uses a composite key: (ID, priority) - ... Where the ID is a UUID and the priority is an integer. I'm trying to perform a range query now: I want all the rows where the ID matches some fixed UUID, but within a range of priorities. This is supported even if I'm using a RandomPartitioner, right? (Because the first key in the composite key is the partition key, and the second part of the composite key is automatically ordered?) So I perform a range slices query: val rangeQuery = HFactory.createRangeSlicesQuery(keyspace, new CompositeSerializer, StringSerializer.get, BytesArraySerializer.get) rangeQuery.setColumnFamily(RouteColumnFamilyName). setKeys( new Composite(id, priorityStart), new Composite(id, priorityEnd) ). setRange( null, null, false, Int.MaxValue ) But I get this error: me.prettyprint.hector.api.exceptions.HInvalidRequestException: InvalidRequestException(why:start key's md5 sorts after end key's md5. this is not allowed; you probably should not specify end key at all, under RandomPartitioner) Shouldn't they have the same md5, since they have the same partition key? Am I using the wrong query here, or does Hector not support composte range queries, or am I making some mistake in how I think Cassandra's composite keys work? Thanks, John
Re: Composite keys and range queries
Ahhh, ok, I thought that CQL was just being brought up to date with the functionality already built into composite keys, but I guess I was mistaken there. But I guess it's just providing a convenient abstraction, using composite column names under the hood. That's where I was confused, thanks. So, in terms of composite column names vs supercolumns: is the only advantage to composite column names that you can do column slicing on subsets of the subcolumns? I.e. if I don't mind loading all of the subcolumns for a given supercolumn name in memory at once (since I need them all anyway), is there any disadvantage to using supercolumns here? They seem a little cleaner and more straightforward for my use case, since I don't have the advantage of the CQL composite key thing. Thanks, John On Wed, Mar 14, 2012 at 12:53 PM, Jeremiah Jordan jeremiah.jor...@morningstar.com wrote: Right, so until the new CQL stuff exists to actually query with something smart enough to know about composite keys , You have to define and query on your own. Row Key = UUID Column = CompositeColumn(string, string) You want to then use COLUMN slicing, not row ranges to query the data. Where you slice in priority as the first part of a Composite Column Name. See the Under the hood and historical notes section of the blog post. You want to layout your data per the Physical representation of the denormalized timeline rows diagram. Where your UUID is the user_id from the example, and your priority is the tweet_id -Jeremiah -- *From:* John Laban [j...@pagerduty.com] *Sent:* Wednesday, March 14, 2012 12:37 PM *To:* user@cassandra.apache.org *Subject:* Re: Composite keys and range queries Hmm, now I'm really confused. This may be of use to you http://www.datastax.com/dev/blog/schema-in-cassandra-1-1 This article is what I actually used to come up with my schema here. In the Clustering, composite keys, and more section they're using a schema very similarly to how I'm trying to use it. They define a composite key with two parts, expecting the first part to be used as the partition key and the second part to be used for ordering. The hash for (uuid-1 , p1) may be 100 and the hash for (uuid-1, p2) may be 1 . Why? Shouldn't only uuid-1 be used as the partition key? (So shouldn't those two hash to the same location?) I'm thinking of using supercolumns for this instead as I know they'll work (where the row key is the uuid and the supercolumn name is the priority), but aren't composite row keys supposed to essentially replace the need for supercolumns? Thanks, and sorry if I'm getting this all wrong, John On Wed, Mar 14, 2012 at 12:52 AM, aaron morton aa...@thelastpickle.comwrote: You are seeing this http://wiki.apache.org/cassandra/FAQ#range_rp The hash for (uuid-1 , p1) may be 100 and the hash for (uuid-1, p2) may be 1 . You cannot do what you want to. Even if you passed a start of (uuid1,empty) and no finish, you would not only get rows where the key starts with uuid1. This may be of use to you http://www.datastax.com/dev/blog/schema-in-cassandra-1-1 Or you can store all the priorities that are valid for an ID in another row. Cheers - Aaron Morton Freelance Developer @aaronmorton http://www.thelastpickle.com On 14/03/2012, at 1:05 PM, John Laban wrote: Forwarding to the Cassandra mailing list as well, in case this is more of an issue on how I'm using Cassandra. Am I correct to assume that I can use range queries on composite row keys, even when using a RandomPartitioner, if I make sure that the first part of the composite key is fixed? Any help would be appreciated, John On Tue, Mar 13, 2012 at 12:15 PM, John Laban j...@pagerduty.com wrote: Hi, I have a column family that uses a composite key: (ID, priority) - ... Where the ID is a UUID and the priority is an integer. I'm trying to perform a range query now: I want all the rows where the ID matches some fixed UUID, but within a range of priorities. This is supported even if I'm using a RandomPartitioner, right? (Because the first key in the composite key is the partition key, and the second part of the composite key is automatically ordered?) So I perform a range slices query: val rangeQuery = HFactory.createRangeSlicesQuery(keyspace, new CompositeSerializer, StringSerializer.get, BytesArraySerializer.get) rangeQuery.setColumnFamily(RouteColumnFamilyName). setKeys( new Composite(id, priorityStart), new Composite(id, priorityEnd) ). setRange( null, null, false, Int.MaxValue ) But I get this error: me.prettyprint.hector.api.exceptions.HInvalidRequestException: InvalidRequestException(why:start key's md5 sorts after end key's md5. this is not allowed; you probably should not specify end key at all, under RandomPartitioner) Shouldn't they have the same md5
Re: Composite keys and range queries
Forwarding to the Cassandra mailing list as well, in case this is more of an issue on how I'm using Cassandra. Am I correct to assume that I can use range queries on composite row keys, even when using a RandomPartitioner, if I make sure that the first part of the composite key is fixed? Any help would be appreciated, John On Tue, Mar 13, 2012 at 12:15 PM, John Laban j...@pagerduty.com wrote: Hi, I have a column family that uses a composite key: (ID, priority) - ... Where the ID is a UUID and the priority is an integer. I'm trying to perform a range query now: I want all the rows where the ID matches some fixed UUID, but within a range of priorities. This is supported even if I'm using a RandomPartitioner, right? (Because the first key in the composite key is the partition key, and the second part of the composite key is automatically ordered?) So I perform a range slices query: val rangeQuery = HFactory.createRangeSlicesQuery(keyspace, new CompositeSerializer, StringSerializer.get, BytesArraySerializer.get) rangeQuery.setColumnFamily(RouteColumnFamilyName). setKeys( new Composite(id, priorityStart), new Composite(id, priorityEnd) ). setRange( null, null, false, Int.MaxValue ) But I get this error: me.prettyprint.hector.api.exceptions.HInvalidRequestException: InvalidRequestException(why:start key's md5 sorts after end key's md5. this is not allowed; you probably should not specify end key at all, under RandomPartitioner) Shouldn't they have the same md5, since they have the same partition key? Am I using the wrong query here, or does Hector not support composte range queries, or am I making some mistake in how I think Cassandra's composite keys work? Thanks, John
Re: best practices for simulating transactions in Cassandra
I'm actually using Curator as a Zookeeper client myself. I haven't used it in production yet, but so far it seems well written and Jordan Zimmerman at Netflix has been great on the support end as well. I haven't tried Cages so I can't really compare, but I think one of the main deciding factors between the two depends on which zk recipes you need. John On Thu, Dec 15, 2011 at 12:07 AM, Boris Yen yulin...@gmail.com wrote: I am not sure if this is the right thread to ask about this. I read that some people are using cage+zookeeper. I was wondering if anyone evaluates https://github.com/Netflix/curator? this seems to be a versatile package. On Tue, Dec 13, 2011 at 6:06 AM, John Laban j...@pagerduty.com wrote: Ok, great. I'll be sure to look into the virtualization-specific NTP guides. Another benefit of using Cassandra over Zookeeper for locking is that you don't have to worry about losing your connection to Zookeeper (and with it your locks) while hammering away at data in Cassandra. If using Cassandra for locks, if you lose your locks you lose your connection to the datastore too. (We're using long-ish session timeouts + connection listeners in ZK to mitigate that now.) John On Mon, Dec 12, 2011 at 12:55 PM, Dominic Williams dwilli...@fightmymonster.com wrote: Hi John, On 12 December 2011 19:35, John Laban j...@pagerduty.com wrote: So I responded to your algorithm in another part of this thread (very interesting) but this part of the paper caught my attention: When client application code releases a lock, that lock must not actually be released for a period equal to one millisecond plus twice the maximum possible drift of the clocks in the client computers accessing the Cassandra databases I've been worried about this, and added some arbitrary delay in the releasing of my locks. But I don't like it as it's (A) an arbitrary value and (B) it will - perhaps greatly - reduce the throughput of the more high-contention areas of my system. To fix (B) I'll probably just have to try to get rid of locks all together in these high-contention areas. To fix (A), I'd need to know what the maximum possible drift of my clocks will be. How did you determine this? What value do you use, out of curiosity? What does the network layout of your client machines look like? (Are any of your hosts geographically separated or all running in the same DC? What's the maximum latency between hosts? etc?) Do you monitor the clock skew on an ongoing basis? Am I worrying too much? If you setup NTP carefully no machine should drift more than 4ms say. I forget where, but you'll find the best documentation on how to make a bullet-proof NTP setup on vendor sites for virtualization software (because virtualization software can cause drift so NTP setup has to be just so) What this means is that, for example, to be really safe when a thread releases a lock you should wait say 9ms. Some points:- -- since the sleep is performed before release, an isolated operation should not be delayed at all -- only a waiting thread or a thread requesting a lock immediately it is released will be delayed, and no extra CPU or memory load is involved -- in practice for the vast majority of application layer data operations this restriction will have no effect on overall performance as experienced by a user, because such operations nearly always read and write to data with limited scope, for example the data of two users involved in some transaction -- the clocks issue does mean that you can't really serialize access to more broadly shared data where more than 5 or 10 such requests are made a second, say, but in reality even if the extra 9ms sleep on release wasn't necessary, variability in database operation execution time (say under load, or when something goes wrong) means trouble might occur serializing with that level of contention So in summary, although this drift thing seems bad at first, partly because it is a new consideration, in practice it's no big deal so long as you look after your clocks (and the main issue to watch out for is when application nodes running on virtualization software, hypervisors et al have setup issues that make their clocks drift under load, and it is a good idea to be wary of that) Best, Dominic Sorry for all the questions but I'm very concerned about this particular problem :) Thanks, John On Mon, Dec 12, 2011 at 4:36 AM, Dominic Williams dwilli...@fightmymonster.com wrote: Hi guys, just thought I'd chip in... Fight My Monster is still using Cages, which is working fine, but... I'm looking at using Cassandra to replace Cages/ZooKeeper(!) There are 2 main reasons:- 1. Although a fast ZooKeeper cluster can handle a lot of load (we aren't getting anywhere near to capacity and we do a *lot* of serialisation) at some point it will be necessary to start hashing lock paths onto separate ZooKeeper clusters
Re: best practices for simulating transactions in Cassandra
transactions are added to locking system in Pelops/Hector/Pycassa, Cassandra will provide better performance than ZooKeeper for storing snapshots, especially as transaction size increases Best, Dominic On 11 December 2011 01:53, Guy Incognito dnd1...@gmail.com wrote: you could try writing with the clock of the initial replay entry? On 06/12/2011 20:26, John Laban wrote: Ah, neat. It is similar to what was proposed in (4) above with adding transactions to Cages, but instead of snapshotting the data to be rolled back (the before data), you snapshot the data to be replayed (the after data). And then later, if you find that the transaction didn't complete, you just keep replaying the transaction until it takes. The part I don't understand with this approach though: how do you ensure that someone else didn't change the data between your initial failed transaction and the later replaying of the transaction? You could get lost writes in that situation. Dominic (in the Cages blog post) explained a workaround with that for his rollback proposal: all subsequent readers or writers of that data would have to check for abandoned transactions and roll them back themselves before they could read the data. I don't think this is possible with the XACT_LOG replay approach in these slides though, based on how the data is indexed (cassandra node token + timeUUID). PS: How are you liking Cages? 2011/12/6 Jérémy SEVELLEC jsevel...@gmail.com Hi John, I had exactly the same reflexions. I'm using zookeeper and cage to lock et isolate. but how to rollback? It's impossible so try replay! the idea is explained in this presentation http://www.slideshare.net/mattdennis/cassandra-data-modeling (starting from slide 24) - insert your whole data into one column - make the job - remove (or expire) your column. if there is a problem during making the job, you keep the possibility to replay and replay and replay (synchronously or in a batch). Regards Jérémy 2011/12/5 John Laban j...@pagerduty.com Hello, I'm building a system using Cassandra as a datastore and I have a few places where I am need of transactions. I'm using ZooKeeper to provide locking when I'm in need of some concurrency control or isolation, so that solves that half of the puzzle. What I need now is to sometimes be able to get atomicity across multiple writes by simulating the begin/rollback/commit abilities of a relational DB. In other words, there are places where I need to perform multiple updates/inserts, and if I fail partway through, I would ideally be able to rollback the partially-applied updates. Now, I *know* this isn't possible with Cassandra. What I'm looking for are all the best practices, or at least tips and tricks, so that I can get around this limitation in Cassandra and still maintain a consistent datastore. (I am using quorum reads/writes so that eventual consistency doesn't kick my ass here as well.) Below are some ideas I've been able to dig up. Please let me know if any of them don't make sense, or if there are better approaches: 1) Updates to a row in a column family are atomic. So try to model your data so that you would only ever need to update a single row in a single CF at once. Essentially, you model your data around transactions. This is tricky but can certainly be done in some situations. 2) If you are only dealing with multiple row *inserts* (and not updates), have one of the rows act as a 'commit' by essentially validating the presence of the other rows. For example, say you were performing an operation where you wanted to create an Account row and 5 User rows all at once (this is an unlikely example, but bear with me). You could insert 5 rows into the Users CF, and then the 1 row into the Accounts CF, which acts as the commit. If something went wrong before the Account could be created, any Users that had been created so far would be orphaned and unusable, as your business logic can ensure that they can't exist without an Account. You could also have an offline cleanup process that swept away orphans. 3) Try to model your updates as idempotent column inserts instead. How do you model updates as inserts? Instead of munging the value directly, you could insert a column containing the operation you want to perform (like +5). It would work kind of like the Consistent Vote Counting implementation: ( https://gist.github.com/41 ). How do you make the inserts idempotent? Make sure the column names correspond to a request ID or some other identifier that would be identical across re-drives of a given (perhaps originally failed) request. This could leave your datastore in a temporarily inconsistent state, but would eventually become consistent after a successful re-drive of the original request. 4) You could take an approach like Dominic Williams proposed with Cages: http://ria101
Re: best practices for simulating transactions in Cassandra
Hi Dominic, So I responded to your algorithm in another part of this thread (very interesting) but this part of the paper caught my attention: When client application code releases a lock, that lock must not actually be released for a period equal to one millisecond plus twice the maximum possible drift of the clocks in the client computers accessing the Cassandra databases I've been worried about this, and added some arbitrary delay in the releasing of my locks. But I don't like it as it's (A) an arbitrary value and (B) it will - perhaps greatly - reduce the throughput of the more high-contention areas of my system. To fix (B) I'll probably just have to try to get rid of locks all together in these high-contention areas. To fix (A), I'd need to know what the maximum possible drift of my clocks will be. How did you determine this? What value do you use, out of curiosity? What does the network layout of your client machines look like? (Are any of your hosts geographically separated or all running in the same DC? What's the maximum latency between hosts? etc?) Do you monitor the clock skew on an ongoing basis? Am I worrying too much? Sorry for all the questions but I'm very concerned about this particular problem :) Thanks, John On Mon, Dec 12, 2011 at 4:36 AM, Dominic Williams dwilli...@fightmymonster.com wrote: Hi guys, just thought I'd chip in... Fight My Monster is still using Cages, which is working fine, but... I'm looking at using Cassandra to replace Cages/ZooKeeper(!) There are 2 main reasons:- 1. Although a fast ZooKeeper cluster can handle a lot of load (we aren't getting anywhere near to capacity and we do a *lot* of serialisation) at some point it will be necessary to start hashing lock paths onto separate ZooKeeper clusters, and I tend to believe that these days you should choose platforms that handle sharding themselves (e.g. choose Cassandra rather than MySQL) 2. Why have more components in your system when you can have less!!! KISS Recently I therefore tried to devise an algorithm which can be used to add a distributed locking layer to clients such as Pelops, Hector, Pycassa etc. There is a doc describing the algorithm, to which may be added an appendix describing a protocol so that locking can be interoperable between the clients. That could be extended to describe a protocol for transactions. Word of warning this is a *beta* algorithm that has only been seen by a select group so far, and therefore not even 100% sure it works but there is a useful general discussion regarding serialization of reads/writes so I include it anyway (and since this algorithm is going to be out there now, if there's anyone out there who fancies doing a Z proof or disproof, that would be fantastic). http://media.fightmymonster.com/Shared/docs/Wait%20Chain%20Algorithm.pdf Final word on this re transactions: if/when transactions are added to locking system in Pelops/Hector/Pycassa, Cassandra will provide better performance than ZooKeeper for storing snapshots, especially as transaction size increases Best, Dominic On 11 December 2011 01:53, Guy Incognito dnd1...@gmail.com wrote: you could try writing with the clock of the initial replay entry? On 06/12/2011 20:26, John Laban wrote: Ah, neat. It is similar to what was proposed in (4) above with adding transactions to Cages, but instead of snapshotting the data to be rolled back (the before data), you snapshot the data to be replayed (the after data). And then later, if you find that the transaction didn't complete, you just keep replaying the transaction until it takes. The part I don't understand with this approach though: how do you ensure that someone else didn't change the data between your initial failed transaction and the later replaying of the transaction? You could get lost writes in that situation. Dominic (in the Cages blog post) explained a workaround with that for his rollback proposal: all subsequent readers or writers of that data would have to check for abandoned transactions and roll them back themselves before they could read the data. I don't think this is possible with the XACT_LOG replay approach in these slides though, based on how the data is indexed (cassandra node token + timeUUID). PS: How are you liking Cages? 2011/12/6 Jérémy SEVELLEC jsevel...@gmail.com Hi John, I had exactly the same reflexions. I'm using zookeeper and cage to lock et isolate. but how to rollback? It's impossible so try replay! the idea is explained in this presentation http://www.slideshare.net/mattdennis/cassandra-data-modeling (starting from slide 24) - insert your whole data into one column - make the job - remove (or expire) your column. if there is a problem during making the job, you keep the possibility to replay and replay and replay (synchronously or in a batch). Regards Jérémy 2011/12/5 John Laban j...@pagerduty.com Hello
Re: best practices for simulating transactions in Cassandra
be fairly simple to use a TTL again to make locks auto expire after N seconds, this would make it more like google chubby. It also allows for bad clients to game the system but that's not something that could be dealt with using authorization apis. For legacy reasons the linked code uses super columns but a regular column family will work just fine. -Jake On Mon, Dec 12, 2011 at 7:36 AM, Dominic Williams dwilli...@fightmymonster.com wrote: Hi guys, just thought I'd chip in... Fight My Monster is still using Cages, which is working fine, but... I'm looking at using Cassandra to replace Cages/ZooKeeper(!) There are 2 main reasons:- 1. Although a fast ZooKeeper cluster can handle a lot of load (we aren't getting anywhere near to capacity and we do a *lot* of serialisation) at some point it will be necessary to start hashing lock paths onto separate ZooKeeper clusters, and I tend to believe that these days you should choose platforms that handle sharding themselves (e.g. choose Cassandra rather than MySQL) 2. Why have more components in your system when you can have less!!! KISS Recently I therefore tried to devise an algorithm which can be used to add a distributed locking layer to clients such as Pelops, Hector, Pycassa etc. There is a doc describing the algorithm, to which may be added an appendix describing a protocol so that locking can be interoperable between the clients. That could be extended to describe a protocol for transactions. Word of warning this is a *beta* algorithm that has only been seen by a select group so far, and therefore not even 100% sure it works but there is a useful general discussion regarding serialization of reads/writes so I include it anyway (and since this algorithm is going to be out there now, if there's anyone out there who fancies doing a Z proof or disproof, that would be fantastic). http://media.fightmymonster.com/Shared/docs/Wait%20Chain%20Algorithm.pdf Final word on this re transactions: if/when transactions are added to locking system in Pelops/Hector/Pycassa, Cassandra will provide better performance than ZooKeeper for storing snapshots, especially as transaction size increases Best, Dominic On 11 December 2011 01:53, Guy Incognito dnd1...@gmail.com wrote: you could try writing with the clock of the initial replay entry? On 06/12/2011 20:26, John Laban wrote: Ah, neat. It is similar to what was proposed in (4) above with adding transactions to Cages, but instead of snapshotting the data to be rolled back (the before data), you snapshot the data to be replayed (the after data). And then later, if you find that the transaction didn't complete, you just keep replaying the transaction until it takes. The part I don't understand with this approach though: how do you ensure that someone else didn't change the data between your initial failed transaction and the later replaying of the transaction? You could get lost writes in that situation. Dominic (in the Cages blog post) explained a workaround with that for his rollback proposal: all subsequent readers or writers of that data would have to check for abandoned transactions and roll them back themselves before they could read the data. I don't think this is possible with the XACT_LOG replay approach in these slides though, based on how the data is indexed (cassandra node token + timeUUID). PS: How are you liking Cages? 2011/12/6 Jérémy SEVELLEC jsevel...@gmail.com Hi John, I had exactly the same reflexions. I'm using zookeeper and cage to lock et isolate. but how to rollback? It's impossible so try replay! the idea is explained in this presentation http://www.slideshare.net/mattdennis/cassandra-data-modeling (starting from slide 24) - insert your whole data into one column - make the job - remove (or expire) your column. if there is a problem during making the job, you keep the possibility to replay and replay and replay (synchronously or in a batch). Regards Jérémy 2011/12/5 John Laban j...@pagerduty.com Hello, I'm building a system using Cassandra as a datastore and I have a few places where I am need of transactions. I'm using ZooKeeper to provide locking when I'm in need of some concurrency control or isolation, so that solves that half of the puzzle. What I need now is to sometimes be able to get atomicity across multiple writes by simulating the begin/rollback/commit abilities of a relational DB. In other words, there are places where I need to perform multiple updates/inserts, and if I fail partway through, I would ideally be able to rollback the partially-applied updates. Now, I *know* this isn't possible with Cassandra. What I'm looking for are all the best practices, or at least tips and tricks, so that I can get around this limitation in Cassandra and still maintain a consistent datastore. (I am using quorum reads/writes so that eventual
Re: best practices for simulating transactions in Cassandra
Ah, neat. It is similar to what was proposed in (4) above with adding transactions to Cages, but instead of snapshotting the data to be rolled back (the before data), you snapshot the data to be replayed (the after data). And then later, if you find that the transaction didn't complete, you just keep replaying the transaction until it takes. The part I don't understand with this approach though: how do you ensure that someone else didn't change the data between your initial failed transaction and the later replaying of the transaction? You could get lost writes in that situation. Dominic (in the Cages blog post) explained a workaround with that for his rollback proposal: all subsequent readers or writers of that data would have to check for abandoned transactions and roll them back themselves before they could read the data. I don't think this is possible with the XACT_LOG replay approach in these slides though, based on how the data is indexed (cassandra node token + timeUUID). PS: How are you liking Cages? 2011/12/6 Jérémy SEVELLEC jsevel...@gmail.com Hi John, I had exactly the same reflexions. I'm using zookeeper and cage to lock et isolate. but how to rollback? It's impossible so try replay! the idea is explained in this presentation http://www.slideshare.net/mattdennis/cassandra-data-modeling (starting from slide 24) - insert your whole data into one column - make the job - remove (or expire) your column. if there is a problem during making the job, you keep the possibility to replay and replay and replay (synchronously or in a batch). Regards Jérémy 2011/12/5 John Laban j...@pagerduty.com Hello, I'm building a system using Cassandra as a datastore and I have a few places where I am need of transactions. I'm using ZooKeeper to provide locking when I'm in need of some concurrency control or isolation, so that solves that half of the puzzle. What I need now is to sometimes be able to get atomicity across multiple writes by simulating the begin/rollback/commit abilities of a relational DB. In other words, there are places where I need to perform multiple updates/inserts, and if I fail partway through, I would ideally be able to rollback the partially-applied updates. Now, I *know* this isn't possible with Cassandra. What I'm looking for are all the best practices, or at least tips and tricks, so that I can get around this limitation in Cassandra and still maintain a consistent datastore. (I am using quorum reads/writes so that eventual consistency doesn't kick my ass here as well.) Below are some ideas I've been able to dig up. Please let me know if any of them don't make sense, or if there are better approaches: 1) Updates to a row in a column family are atomic. So try to model your data so that you would only ever need to update a single row in a single CF at once. Essentially, you model your data around transactions. This is tricky but can certainly be done in some situations. 2) If you are only dealing with multiple row *inserts* (and not updates), have one of the rows act as a 'commit' by essentially validating the presence of the other rows. For example, say you were performing an operation where you wanted to create an Account row and 5 User rows all at once (this is an unlikely example, but bear with me). You could insert 5 rows into the Users CF, and then the 1 row into the Accounts CF, which acts as the commit. If something went wrong before the Account could be created, any Users that had been created so far would be orphaned and unusable, as your business logic can ensure that they can't exist without an Account. You could also have an offline cleanup process that swept away orphans. 3) Try to model your updates as idempotent column inserts instead. How do you model updates as inserts? Instead of munging the value directly, you could insert a column containing the operation you want to perform (like +5). It would work kind of like the Consistent Vote Counting implementation: ( https://gist.github.com/41 ). How do you make the inserts idempotent? Make sure the column names correspond to a request ID or some other identifier that would be identical across re-drives of a given (perhaps originally failed) request. This could leave your datastore in a temporarily inconsistent state, but would eventually become consistent after a successful re-drive of the original request. 4) You could take an approach like Dominic Williams proposed with Cages: http://ria101.wordpress.com/2010/05/12/locking-and-transactions-over-cassandra-using-cages/ The gist is that you snapshot all the original values that you're about to munge somewhere else (in his case, ZooKeeper), make your updates, and then delete the snapshot (and that delete needs to be atomic). If the snapshot data was never deleted, then subsequent accessors (even readers) of the data rows need to do the rollback
best practices for simulating transactions in Cassandra
Hello, I'm building a system using Cassandra as a datastore and I have a few places where I am need of transactions. I'm using ZooKeeper to provide locking when I'm in need of some concurrency control or isolation, so that solves that half of the puzzle. What I need now is to sometimes be able to get atomicity across multiple writes by simulating the begin/rollback/commit abilities of a relational DB. In other words, there are places where I need to perform multiple updates/inserts, and if I fail partway through, I would ideally be able to rollback the partially-applied updates. Now, I *know* this isn't possible with Cassandra. What I'm looking for are all the best practices, or at least tips and tricks, so that I can get around this limitation in Cassandra and still maintain a consistent datastore. (I am using quorum reads/writes so that eventual consistency doesn't kick my ass here as well.) Below are some ideas I've been able to dig up. Please let me know if any of them don't make sense, or if there are better approaches: 1) Updates to a row in a column family are atomic. So try to model your data so that you would only ever need to update a single row in a single CF at once. Essentially, you model your data around transactions. This is tricky but can certainly be done in some situations. 2) If you are only dealing with multiple row *inserts* (and not updates), have one of the rows act as a 'commit' by essentially validating the presence of the other rows. For example, say you were performing an operation where you wanted to create an Account row and 5 User rows all at once (this is an unlikely example, but bear with me). You could insert 5 rows into the Users CF, and then the 1 row into the Accounts CF, which acts as the commit. If something went wrong before the Account could be created, any Users that had been created so far would be orphaned and unusable, as your business logic can ensure that they can't exist without an Account. You could also have an offline cleanup process that swept away orphans. 3) Try to model your updates as idempotent column inserts instead. How do you model updates as inserts? Instead of munging the value directly, you could insert a column containing the operation you want to perform (like +5). It would work kind of like the Consistent Vote Counting implementation: ( https://gist.github.com/41 ). How do you make the inserts idempotent? Make sure the column names correspond to a request ID or some other identifier that would be identical across re-drives of a given (perhaps originally failed) request. This could leave your datastore in a temporarily inconsistent state, but would eventually become consistent after a successful re-drive of the original request. 4) You could take an approach like Dominic Williams proposed with Cages: http://ria101.wordpress.com/2010/05/12/locking-and-transactions-over-cassandra-using-cages/ The gist is that you snapshot all the original values that you're about to munge somewhere else (in his case, ZooKeeper), make your updates, and then delete the snapshot (and that delete needs to be atomic). If the snapshot data was never deleted, then subsequent accessors (even readers) of the data rows need to do the rollback of the previous transaction themselves before they can read/write this data. They do the rollback by just overwriting the current values with what is in the snapshot. It offloads the work of the rollback to the next worker that accesses the data. This approach probably needs an generic/high-level programming layer to handle all of the details and complexity, and it doesn't seem like it was ever added to Cages. Are there other approaches or best practices that I missed? I would be very interested in hearing any opinions from those who have tackled these problems before. Thanks! John