[
https://issues.apache.org/jira/browse/IGNITE-3707?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ksenia Rybakova updated IGNITE-3707:
------------------------------------
Description:
Throughput drops to 0 periodically during failover test. Also lots of "Found
long running transaction" and "Found long running cache future" are in logs at
that moment. This happens right after killed node comes back to the grid. When
this node is killed again performance returns to normal.
Load config:
- Benchmark: CacheRandomOperation
- Operations: put, put_all, get, get_all, invoke, invoke_all, remove,
remove_all, put_if_absent, replace
- Heap: 8Gb for servers, 4Gb for clients
- 20 clients, 5 servers
- Number of caches: 12
- Types of caches (atomicity mode): different (atomic, transactional)
- Types of caches (tiered storage mode): different (onheap without eviction,
onheap with eviction, offheap_tired, offheap_values)
- Types of caches (indexing): different (with and without indexes)
- Types of caches (cache mode): different (partitioned, replicated)
- Backups count: 1
- Preload amount: 0.7M
- Key range: 1.4M
- Failover: 1 server is being restarted every 15 minutes (5 min between stop
and start, 10 min between start and next stop; if several servers are
restarted, step between restarts is 30sec)
- Duration: 1.5h
Yardstick properties file and configs are attached.
Logs can be provided on demand.
was:
Throughput drops to 0 periodically during failover test. Also lots of "Found
long running transaction" and "Found long running cache future" are in logs at
that moment. This happens right after killed node comes back to the grid. When
this node is killed again performance returns to normal.
Load config:
- Benchmark: CacheRandomOperation
- Operations: put, put_all, get, get_all, invoke, invoke_all, remove,
remove_all, put_if_absent, replace
- Heap: 8Gb for servers, 4Gb for clients
- 20 clients, 5 servers
- Number of caches: 12
- Types of caches (atomicity mode): different (atomic, transactional)
- Types of caches (tiered storage mode): different (onheap without eviction,
onheap with eviction, offheap_tired, offheap_values)
- Types of caches (indexing): different (with and without indexes)
- Types of caches (cache mode): different (partitioned, replicated)
- Backups count: 1
- Failover: 1 server is being restarted every 15 minutes (5 min between stop
and start, 10 min between start and next stop; if several servers are
restarted, step between restarts is 30sec)
- Duration: 1.5h
Yardstick properties file and configs are attached.
Logs can be provided on demand.
> Throughput drops to 0 during failover test when killed node returns to the
> grid
> -------------------------------------------------------------------------------
>
> Key: IGNITE-3707
> URL: https://issues.apache.org/jira/browse/IGNITE-3707
> Project: Ignite
> Issue Type: Bug
> Components: general
> Affects Versions: 1.6
> Reporter: Ksenia Rybakova
> Attachments: ignite-base-load-config.xml, run-load.properties,
> run-load.xml
>
>
> Throughput drops to 0 periodically during failover test. Also lots of "Found
> long running transaction" and "Found long running cache future" are in logs
> at that moment. This happens right after killed node comes back to the grid.
> When this node is killed again performance returns to normal.
> Load config:
> - Benchmark: CacheRandomOperation
> - Operations: put, put_all, get, get_all, invoke, invoke_all, remove,
> remove_all, put_if_absent, replace
> - Heap: 8Gb for servers, 4Gb for clients
> - 20 clients, 5 servers
> - Number of caches: 12
> - Types of caches (atomicity mode): different (atomic, transactional)
> - Types of caches (tiered storage mode): different (onheap without eviction,
> onheap with eviction, offheap_tired, offheap_values)
> - Types of caches (indexing): different (with and without indexes)
> - Types of caches (cache mode): different (partitioned, replicated)
> - Backups count: 1
> - Preload amount: 0.7M
> - Key range: 1.4M
> - Failover: 1 server is being restarted every 15 minutes (5 min between stop
> and start, 10 min between start and next stop; if several servers are
> restarted, step between restarts is 30sec)
> - Duration: 1.5h
> Yardstick properties file and configs are attached.
> Logs can be provided on demand.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
