Re: Space Based Purging

2016-09-16 Thread Alain RODRIGUEZ 
Hi,

The only "safe way" to remove data from Cassandra is through tombstones
(TTL / Deletes). I am not sure about the problem you are trying to solve
 here. Maybe could you let us know a bit more about what you are trying to
achieve?

If you are trying to expire old data using a stable % of the disk capacity,
I would calculate used space by using the user mode of the cassandra-stress
tool, using you schema and inserting data. You could then extrapolate to
define the right TTL to use, then make sure to use the appropriate
compaction strategy and compaction options.

More about tombstones and deletes, fwiw:
thelastpickle.com/blog/2016/07/27/about-deletes-and-tombstones.html

---
Alain Rodriguez - @arodream - al...@thelastpickle.com
France

The Last Pickle - Apache Cassandra Consulting
http://www.thelastpickle.com

2016-09-16 12:25 GMT+02:00 INDRANIL BASU :

> Hello All,
>Whats the best way to do data purging in C* column families based on
> size of the table in the disk apart from TTL?
>
>
> *Thanks and regards,*
> *-- IB*
>
>
> --
> *From:* Samba 
> *To:* user@cassandra.apache.org
> *Sent:* Friday, 16 September 2016 2:34 PM
> *Subject:* Re: CASSANDRA-5376: CQL IN clause on last key not working when
> schema includes set,list or map
>
> Thanks Tyler, for identifying that this can be fixed now.
>
> Here is the JIRA ticket : CASSANDRA-12654 :
>
> If this is just removing the now obsolete check, then I hope this makes to
> 3.10 release.
>
> Regards,
> Samba
>
> On Fri, Sep 16, 2016 at 1:33 AM, Tyler Hobbs  wrote:
>
> That ticket was just to improve the error message.  From the comments on
> the ticket:
>
> "Unfortunately, handling collections is slightly harder than what
> CASSANDRA-5230 
> aimed for, because we can't do a name query. So this will have to wait for
> CASSANDRA-4762 . In
> the meantime, we should obviously not throw an assertion error so attaching
> a patch to improve validation."
>
> However, it seems like this would be possible to support in Cassandra
> 3.x.  We probably just need to remove the check and verify that it actually
> works.  Can you open a new JIRA ticket for this?
>
> On Thu, Sep 15, 2016 at 12:49 PM, Samba  wrote:
>
> any update on this issue?
>
> the quoted JIRA issue (CASSANDRA-5376) is resolved as fixed in 1.2.4 but
> it is still not possible (even in 3.7)  to use IN operator in queries that
> fetch collection columns.
>
> is the fix only to report better error message that this is not possible
> or was it fixed then but the issue resurfaced in regression?
>
> could you please confirm one way or the other?
>
> Thanks and Regards,
> Samba
>
>
> On Tue, Sep 6, 2016 at 6:34 PM, Samba  wrote:
>
> Hi,
>
> "CASSANDRA-5376: CQL IN clause on last key not working when schema
> includes set,list or map"
>
> is marked resolved in 1.2.4 but i still see the issue (not an Assertion
> Error, but an query validation message)
>
> was the issue resolved only to report proper error message or was it fixed
> to support retrieving collections when query contains IN clause of
> partition/cluster (last) columns?
>
> If it was fixed properly to support retrieving collections with IN clause,
> then is it a bug in 3.7 release that i get the same message?
>
> Could you please explain, if it not fixed as intended, if there are plans
> to support this in future?
>
> Thanks & Regards,
> Samba
>
>
>
>
>
> --
> Tyler Hobbs
> DataStax 
>
>
>
>
>

Space Based Purging

2016-09-16 Thread INDRANIL BASU 
Hello All,
   Whats the best way to do data purging in C* column families based on size of 
the table in the disk apart from TTL?
  Thanks and regards,-- IB

  From: Samba 
 To: user@cassandra.apache.org 
 Sent: Friday, 16 September 2016 2:34 PM
 Subject: Re: CASSANDRA-5376: CQL IN clause on last key not working when schema 
includes set,list or map
   
Thanks Tyler, for identifying that this can be fixed now.

Here is the JIRA ticket : CASSANDRA-12654 :

If this is just removing the now obsolete check, then I hope this makes to 3.10 
release.

Regards,
Samba

On Fri, Sep 16, 2016 at 1:33 AM, Tyler Hobbs  wrote:

That ticket was just to improve the error message.  From the comments on the 
ticket:

"Unfortunately, handling collections is slightly harder than what  aimed for, 
because we can't do a name query. So this will have to wait for . In the 
meantime, we should obviously not throw an assertion error so attaching a patch 
to improve validation."

However, it seems like this would be possible to support in Cassandra 3.x.  We 
probably just need to remove the check and verify that it actually works.  Can 
you open a new JIRA ticket for this?

On Thu, Sep 15, 2016 at 12:49 PM, Samba  wrote:

any update on this issue? 

the quoted JIRA issue (CASSANDRA-5376) is resolved as fixed in 1.2.4 but it is 
still not possible (even in 3.7)  to use IN operator in queries that fetch 
collection columns.

is the fix only to report better error message that this is not possible or was 
it fixed then but the issue resurfaced in regression?

could you please confirm one way or the other?

Thanks and Regards,
Samba

On Tue, Sep 6, 2016 at 6:34 PM, Samba  wrote:

Hi,

"CASSANDRA-5376: CQL IN clause on last key not working when schema includes 
set,list or map"

is marked resolved in 1.2.4 but i still see the issue (not an Assertion Error, 
but an query validation message)

was the issue resolved only to report proper error message or was it fixed to 
support retrieving collections when query contains IN clause of 
partition/cluster (last) columns?

If it was fixed properly to support retrieving collections with IN clause, then 
is it a bug in 3.7 release that i get the same message?

Could you please explain, if it not fixed as intended, if there are plans to 
support this in future?

Thanks & Regards,
Samba





-- 
Tyler Hobbs
DataStax