Hi,

I am doing a POC on MarkLogic with flexible replication to copy data between 
two clusters. For this, I have been doing some testing and here is an issue I 
encountered.
I have enabled the retry task mentioned here 
https://docs.marklogic.com/guide/flexrep/configuring#id_66155 to run every 
minute with retry seconds min, max are 60 and 300 seconds respectively 
configured on the felx rep domain configuration.

When I stop the target MarkLogic server, make some inserts/updates in the 
source cluster for configured domain, MarkLogic is attempting to replicate them 
to target but fails because it is not able to reach target.
While the retry task, I enabled attempts to try to replicate the document 
again, retry is attempted only 4 times and after 4 failed attempts, there are 
no further retries happening.

>From the documentation, I don't see any limitations on the number of retries 
>it would attempt.

When I look at document properties for the documents that are being attempted 
to be replicated, I am seeing a next-try when the tries count is 1/2/3.

<flexrep:target-id>14035759106745755856</flexrep:target-id>
                
<flexrep:next-try>2018-05-16T11:15:00.927-04:00</flexrep:next-try>
                
<flexrep:last-try>2018-05-16T11:14:00.927-04:00</flexrep:last-try>
                <flexrep:tries>1</flexrep:tries>

But once the tries count reaches 4, next-try element is not showing up

        <flexrep:target-id>14035759106745755856</flexrep:target-id>
        <flexrep:last-try>2018-05-16T11:44:28.245-04:00</flexrep:last-try>
        <flexrep:tries>4</flexrep:tries>

Is there a limitation that MarkLogic enforces but not listed on the 
documentation? In the event of failure of a cluster, the disconnection is going 
to be longer than few minutes and I am trying to figure out how to replicate 
the changes that may have happened when the target is down.

I also observed that while next-try points to certain time like 
2018-05-16T11:15:00.927-04:00 for second attempt here, the actual second time 
is minutes later from this time. I understand that the batch size on domain 
configuration makes a difference here, in my testing, I have only 10 documents 
with changes that need to be replicated and the batch size is configured as 
100. So, it could all be done in one task run but that is not what I am seeing.

Any help is appreciated.


Sudheer




_______________________________________________
General mailing list
[email protected]
Manage your subscription at: 
http://developer.marklogic.com/mailman/listinfo/general

Reply via email to