[
https://issues.apache.org/jira/browse/HBASE-17338?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15891676#comment-15891676
]
Anoop Sam John commented on HBASE-17338:
----------------------------------------
bq.Dumb question. dataSize is KV infrastructure + key content + value +
trailing tags and sequenceid if any? i.e. the whole KV? And CellSize is
infrastructure only or rather key+infrastructure?
Not sure whether used dataSize/cellSize here and there. Its data size only
which is the way as u said above. There is no cellSize as such. There is only
cell heap size which is the heap size (total) occupied by the cell impl object
bq.There is one global threshold whether data is onheap or offheap (I probably
got this wrong?)
Ya there is one global threshold in both cases. Moreover in case of off heap,
there is a heap global threshold also. ie. By def 40% of xmx above which we
will force flush. In case of offheap, we have to check this extra thing also or
else there is possibility of global memstore size getting oversized and GC
impacts/OOME.
bq.We could probably but the direct memory would remain allocated until we
restart.
Sorry for not clear here. I dont mean ON/OFF MSLAB over the RS run time. Now we
have a way to turn it ON at cluster level means at RS level. All the regions in
this RS will use this MSLAB then. (Ya we know some cells wont get copied to
MSLAB - increment/append) My Q was whether there is a way for turning off MSLAB
usage for a specific table. Say in RS MSLAB is ON. But the regions from this
particular table wont use MSLAB at all.
> Treat Cell data size under global memstore heap size only when that Cell can
> not be copied to MSLAB
> ---------------------------------------------------------------------------------------------------
>
> Key: HBASE-17338
> URL: https://issues.apache.org/jira/browse/HBASE-17338
> Project: HBase
> Issue Type: Sub-task
> Components: regionserver
> Affects Versions: 2.0.0
> Reporter: Anoop Sam John
> Assignee: Anoop Sam John
> Fix For: 2.0.0
>
> Attachments: HBASE-17338.patch, HBASE-17338_V2.patch,
> HBASE-17338_V2.patch, HBASE-17338_V4.patch, HBASE-17338_V5.patch
>
>
> We have only data size and heap overhead being tracked globally. Off heap
> memstore works with off heap backed MSLAB pool. But a cell, when added to
> memstore, not always getting copied to MSLAB. Append/Increment ops doing an
> upsert, dont use MSLAB. Also based on the Cell size, we sometimes avoid
> MSLAB copy. But now we track these cell data size also under the global
> memstore data size which indicated off heap size in case of off heap
> memstore. For global checks for flushes (against lower/upper watermark
> levels), we check this size against max off heap memstore size. We do check
> heap overhead against global heap memstore size (Defaults to 40% of xmx) But
> for such cells the data size also should be accounted under the heap overhead.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
