Huge delay reading logs from the source server transfer to attunity server

[Attunity_user]

Hello,

We have been running into issue of Attunity not able to pace with the changes from source (Oracle) to target (Oracle) and ending up error with missing the below error in production.  

I am looking for people's experiences of how this tool is successfully configured from a source that sends massive changes that occur every second.  This includes the server installed on which OS, archive log settings, network settings, etc.

Archived Redo log with the sequence #### does not exist, thread 1.

Captured logs by turning on performance and there is huge latency build up over the time.  The delay is between the source server transferring a file to Attunity server.

[PERFORMANCE ]T: Completed to read from online Redo log 40##### bytes at offset 0000000037281400 with rc 1, read time is 2252 ms, thread '1'

later...

and read time is 53635 ms, thread '1' (oradcdc_redo.c:877)

We are not able to get this work.

Thanks

AU

[stevenguyen]

i remember there was some issue with older kit where it can not read thread 1, so best that you are on the latest service pack 6.4 or latest service pack 6.5

- then if you still have issue, post back.

[Attunity_user]

Hi Stevenguyen,

Thank you for your response. The version we installed is below.

attunity replicate/6.4.0.404

Thanks,

AU

[stevenguyen · Accepted solution]

6.4.404 is SP6 ,, 

current 6.4 is SP17 6.4.0.534, best that you open a support ticket to get the latest version and test with latest version.

[Attunity_user]

Stevenguyen:

As suggested, we installed the latest version, 6.5.0.354 , on all three environments including production.  The task still fails with below error after a day.  The archive set up in the source side is 24 hrs and cannot be extended. But the issue is again latency.  It continuously build up over the time and end up with the below error.

Archived Redo log with the sequence 492617 does not exist, thread 1

We are not sure why the tasks cannot catch up reading logs faster with the pace of transactional changes.

Thanks,

AU

[stevenguyen · Accepted solution]

1. if archive is setup for 24 hours, and error Archived Redo log with the sequence 492617 does not exist, thread 1

- you would need to reload.

2. setup performance for trace to see where the latency is at, source or target.

3. you may want to look at logstream, where target is writing to files on Replicate server.

[Attunity_user]

hi,

Since this failure happens every second day, reload is not a viable solution.

2. setup performance for trace to see where the latency is at, source or target.

Done that, and the performance/latency is on reading the archive logs + transfer the file to the attunity server. We use attunity log reader.

3. you may want to look at logstream, where target is writing to files on Replicate server.

Writing to a file on a replicate server.

Thanks,

AU

[MermaidMe]

Hi Steven,

Would you be able to read the below article, it helped me fix this to an extent by setting  readAheadBlocks=500000 with a retention period of 4hrs on the Oracle source. Again recommend discuss it with any of the technicians in the Qlik Support Team.

https://attunitysupport.qlik.com/s/article/Optimize-read-speed-from-Oracle-ASM

Thank You,

Maneesha T

[Attunity_user]

Hi Maneesha,

Where is this readAheadBlocks=500000 that we can change it in the replication configuration?

[Galit_Spivak]

Hi

readAheadBlocks  is internal Qlik Replicate parameters.

(manage endpoint connection -> your endpoint -> advanced tab -> internal parameters)

Have a good day