Post Reply 
 
Thread Rating:
  • 0 Votes - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
[bug] think decent install
03-02-2018, 10:26 AM (This post was last modified: 03-02-2018 01:55 PM by xwcwt.)
Post: #21
RE: [bug] think decent install
(03-01-2018 06:37 PM)zma Wrote:  @xwcwt: please continue working on making DT work after re-installation following your discussion with Ray in this thread. It is needed any way to support "node reinstallation" no matter whether the 2 TODOs in #19 is implemented or not.

Got it.

If AddClusterUser can work well in a similar 'concurrent run' way as discussed in http://tab.d-thinker.org/showthread.php?...http://tab.d-thinker.org/showthread.php?tid=10823&pid=6599 , then we can safely let think-pre to perform AddClusterUser in each node I think.
Find all posts by this user
Quote this message in a reply
03-02-2018, 01:54 PM
Post: #22
RE: [bug] think decent install
(02-27-2018 05:41 PM)zma Wrote:  P2: DT's persistent memory contains data. If one node is lost, some data are lost. These data are 3 parts:
P2.1 replicated persistent data.
---- DT should be able to fetch from alive nodes. Only after the data are re-written, 3 replica are recovered. So, data will be remain under replicated. If I remember correctly on the details.
P2.1 non-replicated persistent data.
---- unavoidly, there will be data lost.
P3.3 non-persistent data.
---- unavoidly, there will be data lost.

To clarify, do we currently support replicated non-persistent data?

I feel we need change the semantics of the message space system to support container volatility. Of course, we want as little changes as possible.
Find all posts by this user
Quote this message in a reply
03-16-2018, 12:32 PM
Post: #23
RE: [bug] think decent install
(03-02-2018 01:54 PM)lingu Wrote:  
(02-27-2018 05:41 PM)zma Wrote:  P2: DT's persistent memory contains data. If one node is lost, some data are lost. These data are 3 parts:
P2.1 replicated persistent data.
---- DT should be able to fetch from alive nodes. Only after the data are re-written, 3 replica are recovered. So, data will be remain under replicated. If I remember correctly on the details.
P2.1 non-replicated persistent data.
---- unavoidly, there will be data lost.
P3.3 non-persistent data.
---- unavoidly, there will be data lost.

To clarify, do we currently support replicated non-persistent data?

I feel we need change the semantics of the message space system to support container volatility. Of course, we want as little changes as possible.

?
Find all posts by this user
Quote this message in a reply
03-16-2018, 01:15 PM
Post: #24
RE: [bug] think decent install
(03-16-2018 12:32 PM)lingu Wrote:  
(03-02-2018 01:54 PM)lingu Wrote:  
(02-27-2018 05:41 PM)zma Wrote:  P2: DT's persistent memory contains data. If one node is lost, some data are lost. These data are 3 parts:
P2.1 replicated persistent data.
---- DT should be able to fetch from alive nodes. Only after the data are re-written, 3 replica are recovered. So, data will be remain under replicated. If I remember correctly on the details.
P2.1 non-replicated persistent data.
---- unavoidly, there will be data lost.
P3.3 non-persistent data.
---- unavoidly, there will be data lost.

To clarify, do we currently support replicated non-persistent data?

I feel we need change the semantics of the message space system to support container volatility. Of course, we want as little changes as possible.

?

My impression is that we do not have replicated non-persistent range.

I will check this part.
Visit this user's website Find all posts by this user
Quote this message in a reply
03-16-2018, 01:24 PM
Post: #25
RE: [bug] think decent install
(03-16-2018 01:15 PM)zma Wrote:  
(03-16-2018 12:32 PM)lingu Wrote:  
(03-02-2018 01:54 PM)lingu Wrote:  
(02-27-2018 05:41 PM)zma Wrote:  P2: DT's persistent memory contains data. If one node is lost, some data are lost. These data are 3 parts:
P2.1 replicated persistent data.
---- DT should be able to fetch from alive nodes. Only after the data are re-written, 3 replica are recovered. So, data will be remain under replicated. If I remember correctly on the details.
P2.1 non-replicated persistent data.
---- unavoidly, there will be data lost.
P3.3 non-persistent data.
---- unavoidly, there will be data lost.

To clarify, do we currently support replicated non-persistent data?

I feel we need change the semantics of the message space system to support container volatility. Of course, we want as little changes as possible.

?

My impression is that we do not have replicated non-persistent range.

I will check this part.

REP ranges are in persistent memory only at current http://tab.d-thinker.org/showthread.php?tid=2720 .
Visit this user's website Find all posts by this user
Quote this message in a reply
Post Reply 


Forum Jump: