SL21W45: Difference between revisions
Jump to navigation
Jump to search
No edit summary |
No edit summary |
||
Line 13: | Line 13: | ||
*Stasi event generation test & performance impact | *Stasi event generation test & performance impact | ||
**Looks good nearly no load on main thread | **Looks good nearly no load on main thread | ||
**Couchbase storage approach SEEMS to be correct. But not yet fully supported from couchbase: https://forums.couchbase.com/t/scope-collection-limit-performance-impact/32146 | |||
**Writing in async larger json chunks seems efficient | **Writing in async larger json chunks seems efficient | ||
**'''Need to decide if we store StasiData as json or binary at some point.''' | **'''Need to decide if we store StasiData as json or binary at some point.''' | ||
***'''Using kryo should drastically reduce stasidata. But our meta objects need to be very good first''' | ***'''Using kryo should drastically reduce stasidata. But our meta objects need to be very good first''' | ||
***'''Probably not gonna implement StasiData searching via n1ql anyway and do that in the java code as we need to implement it anyway for the lastest 1-5minutes of events''' | ***'''Probably not gonna implement StasiData searching via n1ql anyway and do that in the java code as we need to implement it anyway for the lastest 1-5minutes of events''' | ||
*/stasi querying first implementation | |||
**getting metadata | |||
**'''filter stasidata container to only given flags/params''' | |||
**<br /> | |||
*** | *** |
Revision as of 14:24, 10 November 2021
- Chunkli: Unloading hanging chunk [...] causes deadlock. Need to rework that solution
- BUT...we should not fix it as 1.18 is probably gonna break it anyways. Lets wait and save the time
- BUT...we should not fix it as 1.18 is probably gonna break it anyways. Lets wait and save the time
- Stasi system
- Fully switch to Scopes & Collections
- Remove all Views
- Use water-break, tnt-break
- Do not log UUID of TNTPrimed / ItemSpawn
- Stasi event generation test & performance impact
- Looks good nearly no load on main thread
- Couchbase storage approach SEEMS to be correct. But not yet fully supported from couchbase: https://forums.couchbase.com/t/scope-collection-limit-performance-impact/32146
- Writing in async larger json chunks seems efficient
- Need to decide if we store StasiData as json or binary at some point.
- Using kryo should drastically reduce stasidata. But our meta objects need to be very good first
- Probably not gonna implement StasiData searching via n1ql anyway and do that in the java code as we need to implement it anyway for the lastest 1-5minutes of events
- /stasi querying first implementation
- getting metadata
- filter stasidata container to only given flags/params