[jira] [Updated] (HBASE-9754) Consider eliminating threadlocals from MVCC code
[
https://issues.apache.org/jira/browse/HBASE-9754?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ted Yu updated HBASE-9754:
--
Attachment: 9754-rp-hregion-v3.txt
> Consider eliminating threadlocals from MVCC code
>
>
> Key: HBASE-9754
> URL: https://issues.apache.org/jira/browse/HBASE-9754
> Project: HBase
> Issue Type: Bug
>Reporter: Lars Hofhansl
>Assignee: Ted Yu
> Fix For: 0.98.0
>
> Attachments: 9754-rp-0.txt, 9754-rp-1.txt, 9754-rp-hregion.txt,
> 9754-rp-hregion-v2.txt, 9754-rp-hregion-v3.txt
>
>
> Brought up by [~vrodionov] and [~yuzhih...@gmail.com].
> Currently we use ThreadLocals to communicate the current readpoint between a
> RegionScanner and the Store\{File}Scanner's down the stack.
> Since ThreadLocals are not cheap we should consider whether it is possible to
> pass the readpoint through the call stack instead.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Updated] (HBASE-9754) Consider eliminating threadlocals from MVCC code
[
https://issues.apache.org/jira/browse/HBASE-9754?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ted Yu updated HBASE-9754:
--
Attachment: (was: 9754-rp-hregion-v3.txt)
> Consider eliminating threadlocals from MVCC code
>
>
> Key: HBASE-9754
> URL: https://issues.apache.org/jira/browse/HBASE-9754
> Project: HBase
> Issue Type: Bug
>Reporter: Lars Hofhansl
>Assignee: Ted Yu
> Fix For: 0.98.0
>
> Attachments: 9754-rp-0.txt, 9754-rp-1.txt, 9754-rp-hregion.txt,
> 9754-rp-hregion-v2.txt
>
>
> Brought up by [~vrodionov] and [~yuzhih...@gmail.com].
> Currently we use ThreadLocals to communicate the current readpoint between a
> RegionScanner and the Store\{File}Scanner's down the stack.
> Since ThreadLocals are not cheap we should consider whether it is possible to
> pass the readpoint through the call stack instead.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Updated] (HBASE-9754) Consider eliminating threadlocals from MVCC code
[
https://issues.apache.org/jira/browse/HBASE-9754?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ted Yu updated HBASE-9754:
--
Attachment: 9754-rp-hregion-v3.txt
Updated 9754-rp-hregion-v3.txt with comment on why 0 is passed as readpoint in
certain places.
Also replaced smallestReadPoint in tests.
> Consider eliminating threadlocals from MVCC code
>
>
> Key: HBASE-9754
> URL: https://issues.apache.org/jira/browse/HBASE-9754
> Project: HBase
> Issue Type: Bug
>Reporter: Lars Hofhansl
>Assignee: Ted Yu
> Fix For: 0.98.0
>
> Attachments: 9754-rp-0.txt, 9754-rp-1.txt, 9754-rp-hregion.txt,
> 9754-rp-hregion-v2.txt
>
>
> Brought up by [~vrodionov] and [~yuzhih...@gmail.com].
> Currently we use ThreadLocals to communicate the current readpoint between a
> RegionScanner and the Store\{File}Scanner's down the stack.
> Since ThreadLocals are not cheap we should consider whether it is possible to
> pass the readpoint through the call stack instead.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Updated] (HBASE-9754) Consider eliminating threadlocals from MVCC code
[
https://issues.apache.org/jira/browse/HBASE-9754?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ted Yu updated HBASE-9754:
--
Attachment: (was: 9754-rp-hregion-v3.txt)
> Consider eliminating threadlocals from MVCC code
>
>
> Key: HBASE-9754
> URL: https://issues.apache.org/jira/browse/HBASE-9754
> Project: HBase
> Issue Type: Bug
>Reporter: Lars Hofhansl
>Assignee: Ted Yu
> Fix For: 0.98.0
>
> Attachments: 9754-rp-0.txt, 9754-rp-1.txt, 9754-rp-hregion.txt,
> 9754-rp-hregion-v2.txt
>
>
> Brought up by [~vrodionov] and [~yuzhih...@gmail.com].
> Currently we use ThreadLocals to communicate the current readpoint between a
> RegionScanner and the Store\{File}Scanner's down the stack.
> Since ThreadLocals are not cheap we should consider whether it is possible to
> pass the readpoint through the call stack instead.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Updated] (HBASE-9754) Consider eliminating threadlocals from MVCC code
[
https://issues.apache.org/jira/browse/HBASE-9754?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ted Yu updated HBASE-9754:
--
Attachment: 9754-rp-hregion-v3.txt
New patch that replaces 0 read point parameter value with proper value.
> Consider eliminating threadlocals from MVCC code
>
>
> Key: HBASE-9754
> URL: https://issues.apache.org/jira/browse/HBASE-9754
> Project: HBase
> Issue Type: Bug
>Reporter: Lars Hofhansl
>Assignee: Ted Yu
> Fix For: 0.98.0
>
> Attachments: 9754-rp-0.txt, 9754-rp-1.txt, 9754-rp-hregion.txt,
> 9754-rp-hregion-v2.txt, 9754-rp-hregion-v3.txt
>
>
> Brought up by [~vrodionov] and [~yuzhih...@gmail.com].
> Currently we use ThreadLocals to communicate the current readpoint between a
> RegionScanner and the Store\{File}Scanner's down the stack.
> Since ThreadLocals are not cheap we should consider whether it is possible to
> pass the readpoint through the call stack instead.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Updated] (HBASE-9754) Consider eliminating threadlocals from MVCC code
[
https://issues.apache.org/jira/browse/HBASE-9754?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ted Yu updated HBASE-9754:
--
Attachment: 9754-rp-hregion-v2.txt
Good point, Chunhui.
In latest patch, I refactored Compactor#createFileScanners() and
Compactor#setSmallestReadPoint() which are called by DefaultCompactor#compact().
Now store.getSmallestReadPoint() is only called once.
> Consider eliminating threadlocals from MVCC code
>
>
> Key: HBASE-9754
> URL: https://issues.apache.org/jira/browse/HBASE-9754
> Project: HBase
> Issue Type: Bug
>Reporter: Lars Hofhansl
>Assignee: Ted Yu
> Fix For: 0.98.0
>
> Attachments: 9754-rp-0.txt, 9754-rp-1.txt, 9754-rp-hregion.txt,
> 9754-rp-hregion-v2.txt
>
>
> Brought up by [~vrodionov] and [~yuzhih...@gmail.com].
> Currently we use ThreadLocals to communicate the current readpoint between a
> RegionScanner and the Store\{File}Scanner's down the stack.
> Since ThreadLocals are not cheap we should consider whether it is possible to
> pass the readpoint through the call stack instead.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Updated] (HBASE-9754) Consider eliminating threadlocals from MVCC code
[
https://issues.apache.org/jira/browse/HBASE-9754?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ted Yu updated HBASE-9754:
--
Attachment: 9754-rp-hregion.txt
9754-rp-hregion.txt avoids changing signature of preStoreScannerOpen() by
allowing RegionObserver to query underlying region the correct read point.
> Consider eliminating threadlocals from MVCC code
>
>
> Key: HBASE-9754
> URL: https://issues.apache.org/jira/browse/HBASE-9754
> Project: HBase
> Issue Type: Bug
>Reporter: Lars Hofhansl
>Assignee: Ted Yu
> Fix For: 0.98.0
>
> Attachments: 9754-rp-0.txt, 9754-rp-1.txt, 9754-rp-hregion.txt
>
>
> Brought up by [~vrodionov] and [~yuzhih...@gmail.com].
> Currently we use ThreadLocals to communicate the current readpoint between a
> RegionScanner and the Store\{File}Scanner's down the stack.
> Since ThreadLocals are not cheap we should consider whether it is possible to
> pass the readpoint through the call stack instead.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Updated] (HBASE-9754) Consider eliminating threadlocals from MVCC code
[
https://issues.apache.org/jira/browse/HBASE-9754?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ted Yu updated HBASE-9754:
--
Attachment: 9754-rp-1.txt
New patch that adds readpoint parameter to preStoreScannerOpen().
The reasoning is that when user overrides preStoreScannerOpen(), the user
should know the readpoint which makes store scanner function correctly.
> Consider eliminating threadlocals from MVCC code
>
>
> Key: HBASE-9754
> URL: https://issues.apache.org/jira/browse/HBASE-9754
> Project: HBase
> Issue Type: Bug
>Reporter: Lars Hofhansl
>Assignee: Ted Yu
> Fix For: 0.98.0
>
> Attachments: 9754-rp-0.txt, 9754-rp-1.txt
>
>
> Brought up by [~vrodionov] and [~yuzhih...@gmail.com].
> Currently we use ThreadLocals to communicate the current readpoint between a
> RegionScanner and the Store\{File}Scanner's down the stack.
> Since ThreadLocals are not cheap we should consider whether it is possible to
> pass the readpoint through the call stack instead.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Updated] (HBASE-9754) Consider eliminating threadlocals from MVCC code
[
https://issues.apache.org/jira/browse/HBASE-9754?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ted Yu updated HBASE-9754:
--
Attachment: 9754-rp-0.txt
Rebased patch.
> Consider eliminating threadlocals from MVCC code
>
>
> Key: HBASE-9754
> URL: https://issues.apache.org/jira/browse/HBASE-9754
> Project: HBase
> Issue Type: Bug
>Reporter: Lars Hofhansl
>Assignee: Ted Yu
> Fix For: 0.98.0
>
> Attachments: 9754-rp-0.txt
>
>
> Brought up by [~vrodionov] and [~yuzhih...@gmail.com].
> Currently we use ThreadLocals to communicate the current readpoint between a
> RegionScanner and the Store\{File}Scanner's down the stack.
> Since ThreadLocals are not cheap we should consider whether it is possible to
> pass the readpoint through the call stack instead.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Updated] (HBASE-9754) Consider eliminating threadlocals from MVCC code
[
https://issues.apache.org/jira/browse/HBASE-9754?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ted Yu updated HBASE-9754:
--
Attachment: (was: 9754-rp-0.txt)
> Consider eliminating threadlocals from MVCC code
>
>
> Key: HBASE-9754
> URL: https://issues.apache.org/jira/browse/HBASE-9754
> Project: HBase
> Issue Type: Bug
>Reporter: Lars Hofhansl
>Assignee: Ted Yu
> Fix For: 0.98.0
>
> Attachments: 9754-rp-0.txt
>
>
> Brought up by [~vrodionov] and [~yuzhih...@gmail.com].
> Currently we use ThreadLocals to communicate the current readpoint between a
> RegionScanner and the Store\{File}Scanner's down the stack.
> Since ThreadLocals are not cheap we should consider whether it is possible to
> pass the readpoint through the call stack instead.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Updated] (HBASE-9754) Consider eliminating threadlocals from MVCC code
[
https://issues.apache.org/jira/browse/HBASE-9754?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ted Yu updated HBASE-9754:
--
Attachment: 9754-rp-0.txt
Initial patch.
I ran the following unit tests locally and they passed:
{code}
1164 mt2 -Dtest=TestMemStore
1167 mt2
-Dtest=TestCoprocessorScanPolicy,TestMemStoreChunkPool,TestFSErrorsExposed,TestRegionObserverScannerOpenHook
1173 mt2 -Dtest=TestHRegion
1175 mt2 -Dtest=TestJoinedScanners
{code}
API changes are in classes marked @InterfaceAudience.Private
> Consider eliminating threadlocals from MVCC code
>
>
> Key: HBASE-9754
> URL: https://issues.apache.org/jira/browse/HBASE-9754
> Project: HBase
> Issue Type: Bug
>Reporter: Lars Hofhansl
>Assignee: Ted Yu
> Fix For: 0.98.0
>
> Attachments: 9754-rp-0.txt
>
>
> Brought up by [~vrodionov] and [~yuzhih...@gmail.com].
> Currently we use ThreadLocals to communicate the current readpoint between a
> RegionScanner and the Store\{File}Scanner's down the stack.
> Since ThreadLocals are not cheap we should consider whether it is possible to
> pass the readpoint through the call stack instead.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Updated] (HBASE-9754) Consider eliminating threadlocals from MVCC code
[
https://issues.apache.org/jira/browse/HBASE-9754?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ted Yu updated HBASE-9754:
--
Fix Version/s: 0.98.0
Status: Patch Available (was: Open)
> Consider eliminating threadlocals from MVCC code
>
>
> Key: HBASE-9754
> URL: https://issues.apache.org/jira/browse/HBASE-9754
> Project: HBase
> Issue Type: Bug
>Reporter: Lars Hofhansl
>Assignee: Ted Yu
> Fix For: 0.98.0
>
> Attachments: 9754-rp-0.txt
>
>
> Brought up by [~vrodionov] and [~yuzhih...@gmail.com].
> Currently we use ThreadLocals to communicate the current readpoint between a
> RegionScanner and the Store\{File}Scanner's down the stack.
> Since ThreadLocals are not cheap we should consider whether it is possible to
> pass the readpoint through the call stack instead.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Updated] (HBASE-9754) Consider eliminating threadlocals from MVCC code
[
https://issues.apache.org/jira/browse/HBASE-9754?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Lars Hofhansl updated HBASE-9754:
-
Description:
Brought up by [~vrodionov] and [~yuzhih...@gmail.com].
Currently we use ThreadLocals to communicate the current readpoint between a
RegionScanner and the Store\{File}Scanner's down the stack.
Since ThreadLocals are not cheap we should consider whether it is possible to
pass the readpoint through the call stack instead.
was:
Brought up by [~vrodionov] and [~yuzhih...@gmail.com].
Currently we use ThreadLocals to communicate the current readpoint between a
RegionScanner and the Store{File}Scanner's down the stack.
Since ThreadLocals are not cheap we should consider whether it is possible to
pass the readpoint through the call stack instead.
> Consider eliminating threadlocals from MVCC code
>
>
> Key: HBASE-9754
> URL: https://issues.apache.org/jira/browse/HBASE-9754
> Project: HBase
> Issue Type: Bug
>Reporter: Lars Hofhansl
>
> Brought up by [~vrodionov] and [~yuzhih...@gmail.com].
> Currently we use ThreadLocals to communicate the current readpoint between a
> RegionScanner and the Store\{File}Scanner's down the stack.
> Since ThreadLocals are not cheap we should consider whether it is possible to
> pass the readpoint through the call stack instead.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
