[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15330221#comment-15330221
]
stack commented on HBASE-15594:
---
I am intentionally getting all from cache in these tests [~vrodionov] I am
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15330217#comment-15330217
]
stack commented on HBASE-15594:
---
Yeah. The mismatch between Readers and Handlers via queues costs us big
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15330109#comment-15330109
]
Vladimir Rodionov commented on HBASE-15594:
---
One more thing to consider:
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15321577#comment-15321577
]
Enis Soztutar commented on HBASE-15594:
---
bq. So, again, the Reader doing the whole read/parse of
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15317250#comment-15317250
]
stack commented on HBASE-15594:
---
Yeah. 300k for raw 0.98. Handlers are more occupied, running at about 42
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15317219#comment-15317219
]
stack commented on HBASE-15594:
---
0.98 is faster... 2x? 300k vs 125/140k. Reading from totally cached data
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15316188#comment-15316188
]
Anoop Sam John commented on HBASE-15594:
Interesting Stack.
> [YCSB] Improvements
>
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15316142#comment-15316142
]
stack commented on HBASE-15594:
---
So, again, the Reader doing the whole read/parse of the request and then
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15291547#comment-15291547
]
stack commented on HBASE-15594:
---
bq.. BlockSeek interested to know what you think.
Seems easy enough to
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15291530#comment-15291530
]
ramkrishna.s.vasudevan commented on HBASE-15594:
bq.The blockSeek is interesting as is
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15291495#comment-15291495
]
stack commented on HBASE-15594:
---
Random read, here is where CPU is being spent. We have some work to do:
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15290493#comment-15290493
]
stack commented on HBASE-15594:
---
In YCSB trunk, there is an asynchbase now. I ran some compares. Also tried
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15233177#comment-15233177
]
stack commented on HBASE-15594:
---
There is more to do it seems. I put up
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15229686#comment-15229686
]
stack commented on HBASE-15594:
---
First, thanks for chiming in here [~carp84] Helps to bounce experience off
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15229582#comment-15229582
]
Yu Li commented on HBASE-15594:
---
bq. There is no ConnectionId going on
Oh I meant the below codes in
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15229325#comment-15229325
]
stack commented on HBASE-15594:
---
bq. I'd suggest to remove the hbase.client.ipc.pool.size setting and give
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15229310#comment-15229310
]
stack commented on HBASE-15594:
---
bq. Checking the code, I'm afraid with hbase.client.ipc.pool.size set to
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15228950#comment-15228950
]
stack commented on HBASE-15594:
---
Talking w/ someone versed in this stuff, I should be able to get to 10%
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15228827#comment-15228827
]
stack commented on HBASE-15594:
---
Sorry, I missed this. hyperthreading messes w/ utilization and idle
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15227909#comment-15227909
]
Yu Li commented on HBASE-15594:
---
Yes, YCSB-651 is indeed a pain to find since lots of changes from 0.98.12
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15227798#comment-15227798
]
stack commented on HBASE-15594:
---
Hacking out the scheduler so I run the CallRunner in the Reader thread
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15227763#comment-15227763
]
Duo Zhang commented on HBASE-15594:
---
I do not have the permission to change the BIOS option of a server
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15227753#comment-15227753
]
stack commented on HBASE-15594:
---
Thanks [~Apache9] for chiming in. I don't have easy access to the machines
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15227711#comment-15227711
]
Duo Zhang commented on HBASE-15594:
---
[~stack] I guess your '48 cpus' are 2 * 12 cores with
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15227699#comment-15227699
]
stack commented on HBASE-15594:
---
Thank you for chiming in [~carp84] I have a 0.8.0 SNAPSHOT of YCSB that
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15227680#comment-15227680
]
Yu Li commented on HBASE-15594:
---
Yeah, also got a response from YCSB-681. Thanks for the note Sean.
>
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15227283#comment-15227283
]
Sean Busbey commented on HBASE-15594:
-
The number of connections bug was fixed in
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15227017#comment-15227017
]
stack commented on HBASE-15594:
---
Thanks for pointer [~busbey] ... will do both (one point at other).
>
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15227019#comment-15227019
]
stack commented on HBASE-15594:
---
I'm trying to hit 100% but am unable.
> [YCSB] Improvements
>
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15226971#comment-15226971
]
Yu Li commented on HBASE-15594:
---
Also raised an [issue|https://github.com/brianfrankcooper/YCSB/issues/681]
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15226919#comment-15226919
]
Yu Li commented on HBASE-15594:
---
Which version of YCSB you're running sir [~stack]? We happen to do
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15226789#comment-15226789
]
Enis Soztutar commented on HBASE-15594:
---
Great. Coincidentally, I was running YCSB myself
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15226683#comment-15226683
]
Sean Busbey commented on HBASE-15594:
-
Is the plan for updated docs in our [very-outdated YCSB
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15226675#comment-15226675
]
stack commented on HBASE-15594:
---
I think HBASE-11425 helps us get more work done but the issue here is more
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15226658#comment-15226658
]
Anoop Sam John commented on HBASE-15594:
Just wanted to know whether 11425 work helped wrt more
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15226655#comment-15226655
]
stack commented on HBASE-15594:
---
branch-1. branch-1 because I want to bring fixes into branch-1. You want
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15226634#comment-15226634
]
Anoop Sam John commented on HBASE-15594:
So u are doing this test on trunk or branch-1?
> [YCSB]
[
https://issues.apache.org/jira/browse/HBASE-15594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15226612#comment-15226612
]
stack commented on HBASE-15594:
---
I think this a critical issue given we should be able to use all of a
38 matches
Mail list logo