Re: Read gets stale data after failure of commit phase in CAS operation

[Yuji Ito]

> Can you outline the detailed steps to reproduce the issue Yuji ?
>

Our application requests UPDATE a row.
The row has "id"(PRIMARY KEY), "revision" and "value" columns.
The updating condition is EQ of current "revision".

When the problem happens, the following sequence is executed in one thread.
But a lot of threads request read/write to Cassandra concurrently.
So, all nodes have high load.

1. request updating a row with new revision and value
2. WriteTimeoutException with WriteType.SIMPLE occurs
3. read the row to repair with ConsistencyLevel.SERIAL
4. the read succeeds. But the result isn't new value

This problem does not always occurs.
Almost read can get the latest value.

Thanks
Yuji


On Mon, Jul 25, 2016 at 4:34 PM, DuyHai Doan  wrote:

> Can you outline the detailed steps to reproduce the issue Yuji ?
>
> On Mon, Jul 25, 2016 at 3:16 AM, Yuji Ito  wrote:
>
>> thanks, Ryan
>>
>>
>>> are you using one of the SERIAL Consistency Levels?
>>
>> yes,  I use read with ConsistencyLevel.SERIAL.
>>
>> On Mon, Jul 25, 2016 at 10:10 AM, Ryan Svihla  wrote:
>>
>>> are you using one of the SERIAL Consistency Levels?
>>>
>>> --
>>> Ryan Svihla
>>>
>>> On July 24, 2016 at 8:08:01 PM, Yuji Ito (y...@imagine-orb.com) wrote:
>>>
 Hi,

 I have another question about CAS operation.

 Can a read get stale data after failure in commit phase?

 According to the following article,
 when a write fails in commit phase (a WriteTimeout with WriteType
 SIMPLE happens),
 a subsequent read will repair the uncommitted state
 and get the latest data which the write tried to update.

 Ref.
 http://www.datastax.com/dev/blog/cassandra-error-handling-done-right

 However, when I tested the process,
 the subsequent read can get the previous data which should be
 overwritten.

 Is the repair not always executed in the read?

 Thanks,
 Yuji

>>>
>>
>

Re: Read gets stale data after failure of commit phase in CAS operation

[DuyHai Doan]

Can you outline the detailed steps to reproduce the issue Yuji ?

On Mon, Jul 25, 2016 at 3:16 AM, Yuji Ito  wrote:

> thanks, Ryan
>
>
>> are you using one of the SERIAL Consistency Levels?
>
> yes,  I use read with ConsistencyLevel.SERIAL.
>
> On Mon, Jul 25, 2016 at 10:10 AM, Ryan Svihla  wrote:
>
>> are you using one of the SERIAL Consistency Levels?
>>
>> --
>> Ryan Svihla
>>
>> On July 24, 2016 at 8:08:01 PM, Yuji Ito (y...@imagine-orb.com) wrote:
>>
>>> Hi,
>>>
>>> I have another question about CAS operation.
>>>
>>> Can a read get stale data after failure in commit phase?
>>>
>>> According to the following article,
>>> when a write fails in commit phase (a WriteTimeout with WriteType SIMPLE
>>> happens),
>>> a subsequent read will repair the uncommitted state
>>> and get the latest data which the write tried to update.
>>>
>>> Ref.
>>> http://www.datastax.com/dev/blog/cassandra-error-handling-done-right
>>>
>>> However, when I tested the process,
>>> the subsequent read can get the previous data which should be
>>> overwritten.
>>>
>>> Is the repair not always executed in the read?
>>>
>>> Thanks,
>>> Yuji
>>>
>>
>

Re: Read gets stale data after failure of commit phase in CAS operation

[Yuji Ito]

thanks, Ryan


> are you using one of the SERIAL Consistency Levels?

yes,  I use read with ConsistencyLevel.SERIAL.

On Mon, Jul 25, 2016 at 10:10 AM, Ryan Svihla  wrote:

> are you using one of the SERIAL Consistency Levels?
>
> --
> Ryan Svihla
>
> On July 24, 2016 at 8:08:01 PM, Yuji Ito (y...@imagine-orb.com) wrote:
>
>> Hi,
>>
>> I have another question about CAS operation.
>>
>> Can a read get stale data after failure in commit phase?
>>
>> According to the following article,
>> when a write fails in commit phase (a WriteTimeout with WriteType SIMPLE
>> happens),
>> a subsequent read will repair the uncommitted state
>> and get the latest data which the write tried to update.
>>
>> Ref. http://www.datastax.com/dev/blog/cassandra-error-handling-done-right
>>
>> However, when I tested the process,
>> the subsequent read can get the previous data which should be overwritten.
>>
>> Is the repair not always executed in the read?
>>
>> Thanks,
>> Yuji
>>
>

Re: Read gets stale data after failure of commit phase in CAS operation

[Ryan Svihla]

are you using one of the SERIAL Consistency Levels?

-- 
Ryan Svihla

On July 24, 2016 at 8:08:01 PM, Yuji Ito (y...@imagine-orb.com) wrote:

> Hi,
>
> I have another question about CAS operation.
>
> Can a read get stale data after failure in commit phase?
>
> According to the following article,
> when a write fails in commit phase (a WriteTimeout with WriteType SIMPLE
> happens),
> a subsequent read will repair the uncommitted state
> and get the latest data which the write tried to update.
>
> Ref. http://www.datastax.com/dev/blog/cassandra-error-handling-done-right
>
> However, when I tested the process,
> the subsequent read can get the previous data which should be overwritten.
>
> Is the repair not always executed in the read?
>
> Thanks,
> Yuji
>