[
https://issues.apache.org/jira/browse/IGNITE-8188?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16520266#comment-16520266
]
Amelchev Nikita edited comment on IGNITE-8188 at 6/22/18 11:29 AM:
-------------------------------------------------------------------
[~sergey-chugunov], I have some questions about the issue,
If I understood correctly this methods should need to check request for this
limit and if it is exceeded do nothing? I have looked at using these methods
and when a batch operation fails it processes one-by-one everywhere.
I have suggestions:
1) If the size of the request is exceeded, divide them into valid size batches
and try to process by batches.
2) Make methods "max-size" + "NoNodeException/NodeExistsException" safe. If
batch operation fails then process one-by-one in this methods.
Because request processing one-by-one happens everywhere after failing batch
operation it will allow to avoid code duplication and to close other
issues(IGNITE-8189, TODO in cleanupPreviousClusterData()).
was (Author: nsamelchev):
[sergey-chugunov], I have some questions about the issue,
If I understood correctly this methods should need to check request for this
limit and if it is exceeded do nothing? I have looked at using these methods
and when a batch operation fails it processes one-by-one everywhere.
I have suggestions:
1) If the size of the request is exceeded, divide them into valid size batches
and try to process by batches.
2) Make methods "max-size" + "NoNodeException/NodeExistsException" safe. If
batch operation fails then process one-by-one in this methods.
Because request processing one-by-one happens everywhere after failing batch
operation it will allow to avoid code duplication and to close other
issues(IGNITE-8189, TODO in cleanupPreviousClusterData()).
> Batching operations should perform check for ZooKeeper request max size
> -----------------------------------------------------------------------
>
> Key: IGNITE-8188
> URL: https://issues.apache.org/jira/browse/IGNITE-8188
> Project: Ignite
> Issue Type: Improvement
> Components: zookeeper
> Reporter: Sergey Chugunov
> Assignee: Amelchev Nikita
> Priority: Major
>
> As ZooKeeper documentation
> [says|https://zookeeper.apache.org/doc/r3.4.3/api/org/apache/zookeeper/ZooKeeper.html#multi(java.lang.Iterable)]
> batching *multi* operation has a limit for size of a single request.
> ZookeeperClient batching methods *createAll* and *deleteAll* should check
> this limit and fall back to execute operations one by one.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
