[JBoss JIRA] (ISPN-11128) IndexMode.PRIMARY_OWNER should index data from State Transfer if the index is non-shared
by Gustavo Fernandes (Jira)
[ https://issues.redhat.com/browse/ISPN-11128?page=com.atlassian.jira.plugi... ]
Gustavo Fernandes resolved ISPN-11128.
--------------------------------------
Resolution: Out of Date
After ISPN-11129, non-shared indexes will be able to use IndexMode.ALL
> IndexMode.PRIMARY_OWNER should index data from State Transfer if the index is non-shared
> ----------------------------------------------------------------------------------------
>
> Key: ISPN-11128
> URL: https://issues.redhat.com/browse/ISPN-11128
> Project: Infinispan
> Issue Type: Bug
> Components: Embedded Querying
> Affects Versions: 10.1.0.Final
> Reporter: Gustavo Fernandes
> Assignee: Gustavo Fernandes
> Priority: Major
>
> The {{IndexMode.PRIMARY_OWNER}} strategy is suitable for single indexes cluster-wide, that happens when using {{InfinispanIndexManager}} and {{ElasticSearchIndexManager}}. On this case, an entry should be indexed only once since all new joining nodes will see the global index.
> For the cases where the index is local to each node (near-real-time and file system index managers), the state transfer data should definitely be indexed, otherwise when ownership of a key changes the entry will be kept in the 'old' owner.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
6 years, 1 month
[JBoss JIRA] (ISPN-11299) Stale values can be indexed during State Transfer
by Gustavo Fernandes (Jira)
[ https://issues.redhat.com/browse/ISPN-11299?page=com.atlassian.jira.plugi... ]
Gustavo Fernandes reassigned ISPN-11299:
----------------------------------------
Assignee: Gustavo Fernandes
> Stale values can be indexed during State Transfer
> -------------------------------------------------
>
> Key: ISPN-11299
> URL: https://issues.redhat.com/browse/ISPN-11299
> Project: Infinispan
> Issue Type: Bug
> Components: Core
> Affects Versions: 10.1.1.Final
> Reporter: Gustavo Fernandes
> Assignee: Gustavo Fernandes
> Priority: Major
>
> If an operation is discarded during state transfer because it was already modified locally, the QueryInterceptor still propagates it to the index, leaving the index out of sync.
> Sequence of events:
> * T1: State Transfer starts
> * T2: The EWI (EntryWrappingInterceptor) starts tracking all non-state-transfer operations
> * T3: An entry is added locally
> * T4: EWI stores the key
> * T5: The same key arrives from State Transfer
> * T6: QueryInterceptor indexes it (The QI is installed after EWI but indexing happens before storing the entry in the data container)
> * T7: The entry operation is not committed in the data container since it was tracked before as DiscardPolicy{discardStateTransfer=true, discardXSiteStateTransfer=false}
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
6 years, 1 month
[JBoss JIRA] (ISPN-11303) Local non-blocking get breaks hit/miss statistics
by Dan Berindei (Jira)
Dan Berindei created ISPN-11303:
-----------------------------------
Summary: Local non-blocking get breaks hit/miss statistics
Key: ISPN-11303
URL: https://issues.redhat.com/browse/ISPN-11303
Project: Infinispan
Issue Type: Bug
Components: Hot Rod, Server
Affects Versions: 9.4.17.Final
Reporter: Dan Berindei
Assignee: Dan Berindei
Fix For: 9.4.18.Final
The local-non-blocking get was breaking the hit/miss statistics
when client requests didn't go to the primary owner,
e.g. with a load balancer between the clients and the servers.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
6 years, 2 months
[JBoss JIRA] (ISPN-11129) High Availability for non-shared indexes on DIST caches
by Tristan Tarrant (Jira)
[ https://issues.redhat.com/browse/ISPN-11129?page=com.atlassian.jira.plugi... ]
Tristan Tarrant updated ISPN-11129:
-----------------------------------
Status: Resolved (was: Pull Request Sent)
Resolution: Done
> High Availability for non-shared indexes on DIST caches
> -------------------------------------------------------
>
> Key: ISPN-11129
> URL: https://issues.redhat.com/browse/ISPN-11129
> Project: Infinispan
> Issue Type: Bug
> Components: Embedded Querying
> Affects Versions: 9.4.17.Final, 10.1.1.Final
> Reporter: Gustavo Fernandes
> Assignee: Gustavo Fernandes
> Priority: Major
> Fix For: 10.1.2.Final, 11.0.0.Final
>
>
> 'Non-shared indexes' is a setup when each node has a local index backed by the file system, using either the near-real-time or the FS index managers.
> For REPL caches, a full copy of the index is kept in each node. As long as at least on node is up the queries hit the local index which represents the full index of the data.
> But non-shared indexes have some drawbacks for DIST caches: each node only indexes data that the node is the primary owner. If a node goes down, there is no redundancy. This issue aims to solve this issue at two levels:
> * Indexing: replicated entries should be indexed for redundancy purposes. If a primary owner goes down, the entry is indexed in the backup owner. This should follow exactly the data distribution on the cache in terms of number of replicas and rebalancing
> * Querying: query broadcast is currently used for non-shared indexes. If replicated entries start to be indexed, a mechanism will have to be used to avoid retuning repeated results, since currently the broadcast simply queries all the entries in each node and join the result.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
6 years, 2 months