[
https://issues.apache.org/jira/browse/IGNITE-1494?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Semen Boikov updated IGNITE-1494:
---------------------------------
Priority: Major (was: Critical)
Fix Version/s: (was: ignite-1.4)
ignite-1.5
I found that test can fail only with ATOMIC cache with
atomicWriteOrderMode=CLOCK and only if cache.removeAll() (without parameters)
operation is used.
In CLOCK mode update version is generated on the node initiated update,
cache.removeAll() is implemented as compute job which is executed on server
node, so it is possible that if client executes 'put' right after 'removeAll()'
operation then version generated on client for 'put' will be smaller than
version generated on server for 'remove' and 'put' will be ignored.
Need discuss if it makes sense to fix this issue, for now I suggest to change
test to add 1sec delay after 'removeAll()'.
> Get return nulls for some cache configurations
> ----------------------------------------------
>
> Key: IGNITE-1494
> URL: https://issues.apache.org/jira/browse/IGNITE-1494
> Project: Ignite
> Issue Type: Bug
> Components: cache
> Affects Versions: ignite-1.4
> Reporter: Sergey Kozlov
> Assignee: Semen Boikov
> Fix For: ignite-1.5
>
> Attachments: CacheTestExample.java, grid_config_01.client.xml,
> grid_config_01.xml
>
>
> 1. Copy grid_config_01.xml and grid_config_01.client.xml in examples/config
> directory (it contains 25 FULL_SYNCed cache configurations)
> 2. Copy CacheTestExample in org.apache.ignite.examples.datagrid package
> 3. Start two nodes by bin/ignite.sh examples/config/grid_config_01.xml
> 4. Build and run CacheTestExample (it makes put/get/remove operations for 100
> keys) a few times (w/o stopping server nodes)
> 5. The normal output should be following:
> {noformat}
> [12:12:02] To start Console Management & Monitoring run
> ignitevisorcmd.{sh|bat}
> [12:12:02]
> [12:12:02] Ignite node started OK (id=3f1ea113)
> [12:12:02] Topology snapshot [ver=5, servers=2, clients=1, CPUs=8, heap=3.8GB]
> >>>cache_0001
> Put 100 -> put1_
> Get 100
> RemoveAll 100
> Put 1..50 -> put2_
> Remove 30.40
> RemoveAll 21..50
> PutAll 21..100 -> putall_
> Get 1..20
> Get 21..100
> >>>cache_0002
> Put 100 -> put1_
> Get 100
> RemoveAll 100
> Put 1..50 -> put2_
> Remove 30.40
> RemoveAll 21..50
> PutAll 21..100 -> putall_
> Get 1..20
> Get 21..100
> >>>cache_0003
> Put 100 -> put1_
> Get 100
> RemoveAll 100
> Put 1..50 -> put2_
> Remove 30.40
> RemoveAll 21..50
> PutAll 21..100 -> putall_
> Get 1..20
> Get 21..100
> ...
> {noformat}
> But for some caches I got:
> {noformat}
> >>>cache_0011
> Put 100 -> put1_
> Get 100
> RemoveAll 100
> Put 1..50 -> put2_
> Remove 30.40
> RemoveAll 21..50
> PutAll 21..100 -> putall_
> Get 1..20
> Wrong value found [key=1, val=null]
> Wrong value found [key=2, val=null]
> Wrong value found [key=3, val=null]
> Wrong value found [key=4, val=null]
> Get 21..100
> ...
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
