[ 
https://issues.apache.org/jira/browse/HDDS-7845?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Hemant Kumar updated HDDS-7845:
-------------------------------
    Description: There is an issue while running a UT that included a snapdiff 
call, The problem was that it included 2 consecutive createSnapshot calls , The 
first one created a checkpoint however the second didn’t and gave an error 
during snapshot read (as checkpoint hadn't been created). although it didn’t 
give an error during snapshot create and returned a response, It looks like the 
checkpoint creation is happening async and not blocking call. One more 
observation is that this occurs only when ratisEnabled set to true .   (was: 
Currently, `RocksDBCheckpointDiffer` does lots of things (e.g. maintaining the 
DAG, implements RocksDB listener, diff two snapshots and probably remove/update 
of snapshot from DAG too) which makes it complicated and harder to test because 
of tight coupling with RocksDB.
`RocksDbCheckpointDiffer` can be simplified by extracting out the DAG to new 
class `RocksDbCompactionDag` which maintains SST DAG in which you can add nodes 
& arcs, load the DAG from disk/DB, remove nodes and arcs, etc. DAG is 
independent of what we use to store compaction log (disk or RocksDB). We should 
be able to test all the functionality of `RocksDbCompactionDag` independently.)

> Wait for checkpoint directory to be created
> -------------------------------------------
>
>                 Key: HDDS-7845
>                 URL: https://issues.apache.org/jira/browse/HDDS-7845
>             Project: Apache Ozone
>          Issue Type: Sub-task
>            Reporter: Hemant Kumar
>            Assignee: Hemant Kumar
>            Priority: Major
>
> There is an issue while running a UT that included a snapdiff call, The 
> problem was that it included 2 consecutive createSnapshot calls , The first 
> one created a checkpoint however the second didn’t and gave an error during 
> snapshot read (as checkpoint hadn't been created). although it didn’t give an 
> error during snapshot create and returned a response, It looks like the 
> checkpoint creation is happening async and not blocking call. One more 
> observation is that this occurs only when ratisEnabled set to true . 



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to