Replicate Error

[wert1]

Hi,

I newbie to Qlik, would need  some assistance / guidance in resolving the below error.

: 2023-10-27T08:54:40 [TARGET_APPLY ]E: Failed (retcode -1) to execute statement: 'describe formatted `oms_raw`.`C_DATA__ct`' [1022502] (ar_odbc_stmt.c:4645)
: 2023-10-27T08:54:40 [TARGET_APPLY ]E: RetCode: SQL_ERROR SqlState: 08S01 NativeError: 115 Message: [Cloudera][Hardy] (115) Connection failed with error: Could not refill buffer [1022502] (ar_odbc_stmt.c:4649)
: 2023-10-27T08:54:40 [TARGET_APPLY ]E: Failed to describe table using ODBC. [1022506] (hadoop_metadata.c:436)
: 2023-10-27T08:54:40 [TARGET_APPLY ]E: Failed to get table definition. [1022506] (hadoop_metadata.c:646)
: 2023-10-27T08:54:40 [TARGET_APPLY ]E: Failed to get extended Hortonworks Data Platform (HDP) table def, table name 'C_DATA__ct'. [1022506] (hadoop_apply.c:1437)
: 2023-10-27T08:54:40 [TARGET_APPLY ]E: Failed to create table writer. [1022506] (hadoop_apply.c:1544)
: 2023-10-27T08:54:40 [TARGET_APPLY ]I: Error executing data handler [1022506] (streamcomponent.c:2001)
: 2023-10-27T08:54:40 [TASK_MANAGER ]I: Stream component failed at subtask 0, component st_0_PROD CDP [1022506] (subtask.c:1396)
: 2023-10-27T08:54:40 [TARGET_APPLY ]I: Target component st_0_PROD CDP was detached because of recoverable error. Will try to reattach (subtask.c:1511)

Thanks 

[OritA]

Hi, 

Could you please open a case for your problem and attache to the case the task diagnostic package with the log with the error for further troubleshooting the problem. 

Thanks & regards,

Orit

[Heinvandenheuvel]

Hmm, Newby to Qlik Replicate and newbie to forum article writing  apparently.

"Replicate error" does not help anyone to find it back later, or be enticed to help because they have seen something like that.

Next time pick a subject like "[Cloudera][Hardy] (115) Connection failed with error: Could not refill buffer" or "Failed to get extended Hortonworks Data Platform (HDP) table def"

This FEELS like a drive/version mismatch - please doublecheck all drivers, db version against the Replicate supported versions, and make sure all other prerequisites are matched.

Now for the problem description:

- Did it ever work? (DEV, QA, PROD)

- Do other tasks/tables work?

- Could this be a special table (extra wide?)

- Does a 'test' for the target work?

- Does 'apply' work vs 'store' (if that's an option for the endpoint

- Are there further error details in some target db error log?

Good luck,

Hein.

[wert1]

- Did it ever work? (DEV, QA, PROD)
Yes, we randonmly see this error

- Do other tasks/tables work?
Yes

- Could this be a special table (extra wide?)
Unsure

- Does a 'test' for the target work?

- Does 'apply' work vs 'store' (if that's an option for the endpoint

- Are there further error details in some target db error log?
00003524: 2023-10-27T08:54:58 [TARGET_APPLY ]I: Target components st_0_PROD CDP was reattached after 17 seconds. (subtask.c:1380)

[Heinvandenheuvel]

Based on you answers so far, my guess is that the failing 'describe' statement may well be a victim, not a cause. That there is some sort of intermittent connectivity issues (DNS?) which happened to strike during the request to describe and resolved later.

- is it always about this table and this statement or are there other issues with "Connection failed with error: Could not refill buffer " 

- always, and only once, when the task is starting, or 'hours' into running, and perhaps multiple times? 

- does it looks like processing indeed properly continues after "rget components st_0_PROD CDP was reattached after 17 seconds."

- it 'feels' like some low-memory issue. Maybe some buffer size can be increased - but I wouldn't know what or where.

Hein.

[wert1]

@Heinvandenheuvel  Thanks for your time, please find my reply

- is it always about this table and this statement or are there other issues with "Connection failed with error: Could not refill buffer "
There are other tables and statemnt where we see same error (Failed to execute statement ALTER TABLE `one_staging`.`ACT_AUDIT)

- always, and only once, when the task is starting, or 'hours' into running, and perhaps multiple times?
I think its mostly when the task is starting.

- does it looks like processing indeed properly continues after "rget components st_0_PROD CDP was reattached after 17 seconds."

Once it is reattached , it starts processing witout any issues.

- it 'feels' like some low-memory issue. Maybe some buffer size can be increased - but I wouldn't know what or where.
Is there a way this can be checked ?

Wert

[Heinvandenheuvel · Accepted solution]

>> Once it is reattached , it starts processing without any issues.

So basically this tells us that Replicate is working fine, more than fine as it even recovers from environment errors.

You should start looking at the database server error logs (server, service, network, database) and such .Did a slave process die around the error time?   If need be engage your network team. Sometimes you need to go as deep as 'Wireshark' to look at the message exchanges around the trouble time to see a disconnect 

Hein.