do the size of your rows vary a lot?  if so, then this could explain why some nodes are "unlucky" and getting the large nodes.  you only have 3 nodes so definitely can happen.


On Thu, Oct 11, 2012 at 2:24 PM, Alain RODRIGUEZ <arodrime@gmail.com> wrote:
I understood it that way too but I explained myself very poorly in my last message. Answering fast and not in my mother tongue is sometimes quite hard and results on poor explanations.

Sorry about that and thank you Jim for that clarification.

Alain

2012/10/11 Jim Cistaro <jcistaro@netflix.com>
I think bad effect #1 needs clarification.

This only suspends minor compactions involving that 1 big file.  As new sstables are flushed, they are of the same small size and they will eventually compact together.  So that one big file will sit idle (as far as compaction go) until you build some eventualluy compact the new files into ones as big as tht major compacted sstable.  That is why those who do major compactions once normally end up doing them periodically to cause compaction with the previous large sstable.

I hope this helps,
jc

From: Alain RODRIGUEZ <arodrime@gmail.com>
Reply-To: <user@cassandra.apache.org>
Date: Thu, 11 Oct 2012 17:30:53 +0200

To: <user@cassandra.apache.org>
Subject: Re: unbalanced ring

@Tamar

Bad effects are : 

1 - Disabling for a *long* time your minor compactions (they need SSTable about the same size to be triggered).
2 - High cpu load during the compaction (which can be quite long).

Good effects :

1 - Reduce the size of your data.
2 - Boost read performances.

@B. Todd Burruss

What can we do if cleanup doesn't remove any data and so doesn't balance the data partition ?
We both have well balanced ring and unbalanced data...

Alain

2012/10/11 Viktor Jevdokimov <Viktor.Jevdokimov@adform.com>

In our case, we use TTL and need to keep amount of data as low as possible to fit RAM, so data have to be deleted somehow.

While SSTables are growing, largest will wait long time for minor compaction, so we do major compaction every night.

 

 

Best regards / Pagarbiai
Viktor Jevdokimov
Senior Developer

J. Jasinskio 16C, LT-01112 Vilnius, Lithuania
Follow us on Twitter: @adforminsider
Adform News
   
Visit us at IAB RTB workshop
October 11, 4 pm in Sala Rossa
iab forum

Disclaimer: The information contained in this message and attachments is intended solely for the attention and use of the named addressee and may be confidential. If you are not the intended recipient, you are reminded that the information remains the property of the sender. You must not use, disclose, distribute, copy, print or rely on this e-mail. If you have received this message in error, please contact the sender immediately and irrevocably delete this message and any copies.

From: Tamar Fraenkel [mailto:tamar@tok-media.com]
Sent: Thursday, October 11, 2012 10:57


To: user@cassandra.apache.org
Subject: Re: unbalanced ring

 

Hi!
All that left me confused...
like Alain, I read DataStax wanings. Now Victor says it is possible without bad effects.
1. Under what conditions would you recommend major compaction?
2. If I do go that route, would I have to run periodic / nightly compactions from now on?
3. What will be the price of #2?
Thanks,

Tamar Fraenkel 
Senior Software Engineer, TOK Media 

 

 



On Thu, Oct 11, 2012 at 8:41 AM, Viktor Jevdokimov <Viktor.Jevdokimov@adform.com> wrote:

To run, or not to run? All this depends on use case. There’re no problems running major compactions (we do it nightly) in one case, there could be problems in another. Just need to understand, how everything works.

 

 

Best regards / Pagarbiai

Viktor Jevdokimov

Senior Developer

 

J. Jasinskio 16C, LT-01112 Vilnius, Lithuania

Follow us on Twitter: @adforminsider

What is Adform: watch this short video

Adform News

   

Visit us at IAB RTB workshop

October 11, 4 pm in Sala Rossa

iab forum


Disclaimer: The information contained in this message and attachments is intended solely for the attention and use of the named addressee and may be confidential. If you are not the intended recipient, you are reminded that the information remains the property of the sender. You must not use, disclose, distribute, copy, print or rely on this e-mail. If you have received this message in error, please contact the sender immediately and irrevocably delete this message and any copies.

 

From: Alain RODRIGUEZ [mailto:arodrime@gmail.com]
Sent: Thursday, October 11, 2012 09:17
To: user@cassandra.apache.org
Subject: Re: unbalanced ring

 

Tamar be carefull. Datastax doesn't recommand major compactions in production environnement.

 

If I got it right, performing major compaction will convert all your SSTables into a big one, improving substantially your reads performence, at least for a while... The problem is that will disable minor compactions too (because of the difference of size between this SSTable and the new ones, if I remeber well). So your reads performance will decrease until your others SSTable reach the size of this big one you've created or until you run an other major compaction, transforming them into a maintenance normal process like repair is.

 

But, knowing that, I still don't know if we both (Tamar and I) shouldn't run it anyway (In my case it will greatly decrease the size of my data  133 GB -> 35GB and maybe load the cluster evenly...)

 

Alain

 

2012/10/10 B. Todd Burruss <btoddb@gmail.com>

it should not have any other impact except increased usage of system resources.

 

and i suppose, cleanup would not have an affect (over normal compaction) if all nodes contain the same data

 

On Wed, Oct 10, 2012 at 12:12 PM, Tamar Fraenkel <tamar@tok-media.com> wrote:

Hi!
Apart from being heavy load (the compact), will it have other effects?
Also, will cleanup help if I have replication factor = number of nodes?
Thanks


Tamar Fraenkel 
Senior Software Engineer, TOK Media 

 

 

 

On Wed, Oct 10, 2012 at 6:12 PM, B. Todd Burruss <btoddb@gmail.com> wrote:

major compaction in production is fine, however it is a heavy operation on the node and will take I/O and some CPU.

 

the only time i have seen this happen is when i have changed the tokens in the ring, like "nodetool movetoken".  cassandra does not auto-delete data that it doesn't use anymore just in case you want to move the tokens again or otherwise "undo".

 

try "nodetool cleanup"

 

On Wed, Oct 10, 2012 at 2:01 AM, Alain RODRIGUEZ <arodrime@gmail.com> wrote:

Hi,

 

Same thing here: 

 

2 nodes, RF = 2. RCL = 1, WCL = 1.

Like Tamar I never ran a major compaction and repair once a week each node.

 

10.59.21.241    eu-west     1b          Up     Normal  133.02 GB       50.00%              0

10.58.83.109    eu-west     1b          Up     Normal  98.12 GB        50.00%              85070591730234615865843651857942052864

 

What phenomena could explain the result above ?

 

By the way, I have copy the data and import it in a one node dev cluster. There I have run a major compaction and the size of my data has been significantly reduced (to about 32 GB instead of 133 GB). 

 

How is that possible ?

Do you think that if I run major compaction in both nodes it will balance the load evenly ?

Should I run major compaction in production ?

 

2012/10/10 Tamar Fraenkel <tamar@tok-media.com>

Hi!
I am re-posting this, now that I have more data and still unbalanced ring:

3 nodes,
RF=3, RCL=WCL=QUORUM



Address         DC          Rack        Status State   Load            Owns    Token
                                                                               113427455640312821154458202477256070485

x.x.x.x    us-east     1c          Up     Normal  24.02 GB        33.33%  0
y.y.y.y     us-east     1c          Up     Normal  33.45 GB        33.33%  56713727820156410577229101238628035242
z.z.z.z    us-east     1c          Up     Normal  29.85 GB        33.33%  113427455640312821154458202477256070485

repair runs weekly.
I don't run nodetool compact as I read that this may cause the minor regular compactions not to run and then I will have to run compact manually. Is that right?

Any idea if this means something wrong, and if so, how to solve?



Thanks,


Tamar Fraenkel 

Senior Software Engineer, TOK Media 

 

 

 

On Tue, Mar 27, 2012 at 9:12 AM, Tamar Fraenkel <tamar@tok-media.com> wrote:

Thanks, I will wait and see as data accumulates.

Thanks,


Tamar Fraenkel 
Senior Software Engineer, TOK Media 

 

 

 

On Tue, Mar 27, 2012 at 9:00 AM, R. Verlangen <robin@us2.nl> wrote:

Cassandra is built to store tons and tons of data. In my opinion roughly ~ 6MB per node is not enough data to allow it to become a fully balanced cluster.

 

2012/3/27 Tamar Fraenkel <tamar@tok-media.com>

This morning I have

 nodetool ring -h localhost

Address         DC          Rack        Status State   Load            Owns    Token

                                                                               113427455640312821154458202477256070485

10.34.158.33    us-east     1c          Up     Normal  5.78 MB         33.33%  0

10.38.175.131   us-east     1c          Up     Normal  7.23 MB         33.33%  56713727820156410577229101238628035242

10.116.83.10    us-east     1c          Up     Normal  5.02 MB         33.33%  113427455640312821154458202477256070485

 

Version is 1.0.8.

 

 

Tamar Fraenkel 
Senior Software Engineer, TOK Media 

 

 

 

On Tue, Mar 27, 2012 at 4:05 AM, Maki Watanabe <watanabe.maki@gmail.com> wrote:

What version are you using?

Anyway try nodetool repair & compact.

 

maki

 

2012/3/26 Tamar Fraenkel <tamar@tok-media.com>

Hi!

I created Amazon ring using datastax image and started filling the db.

The cluster seems un-balanced.

 

nodetool ring returns:

Address         DC          Rack        Status State   Load            Owns    Token

                                                                               113427455640312821154458202477256070485

10.34.158.33    us-east     1c          Up     Normal  514.29 KB       33.33%  0

10.38.175.131   us-east     1c          Up     Normal  1.5 MB          33.33%  56713727820156410577229101238628035242

10.116.83.10    us-east     1c          Up     Normal  1.5 MB          33.33%  113427455640312821154458202477256070485

 

[default@tok] describe;

Keyspace: tok:

  Replication Strategy: org.apache.cassandra.locator.SimpleStrategy

  Durable Writes: true

    Options: [replication_factor:2]

 

[default@tok] describe cluster;

Cluster Information:

   Snitch: org.apache.cassandra.locator.Ec2Snitch

   Partitioner: org.apache.cassandra.dht.RandomPartitioner

   Schema versions:

        4687d620-7664-11e1-0000-1bcb936807ff: [10.38.175.131, 10.34.158.33, 10.116.83.10]

 

 

Any idea what is the cause?

I am running similar code on local ring and it is balanced.

 

How can I fix this?

 

Thanks,


Tamar Fraenkel 
Senior Software Engineer, TOK Media 

 

 

 

 

 



 

--
With kind regards,

 

Robin Verlangen