mesos-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dominic Hamon <dha...@twopensource.com>
Subject Twitter sprint Q3.6
Date Mon, 22 Sep 2014 18:41:47 GMT
Hello

Here's our focus for the next two weeks. Some of these we are shepherding,
but are high enough profile that we want to track them.

Link to active sprint
<https://issues.apache.org/jira/secure/RapidBoard.jspa?rapidView=35>

Task reconciliation:
  - Removing out-of-order updates
Isolation:
  - /tmp isolation
  - PID namespaces
  - Expose RTT in container stats
Persistence:
  - Responding to design comments
Reliability:
  - Handling temporary one-way network splits
  - Access to stats.json after framework stops
  - Allow slave reconfiguration on restart


On Tue, Sep 9, 2014 at 10:16 AM, Dominic Hamon <dhamon@twopensource.com>
wrote:

> Hello Mesos developers!
>
> I thought it might be useful to have a little insight into how we work on
> Mesos at Twitter and, more specifically, what we are working on. We work in
> biweekly sprints and starting with this one (that just began yesterday) I'm
> going to start sending out a summary of what we're planning.
>
> I hope you find it useful; feedback is encouraged.
>
> Link to active sprint
> <https://issues.apache.org/jira/secure/RapidBoard.jspa?rapidView=35>
>
> Task reconciliation:
>   - extending master state to include terminal unacknowledged tasks
>   - adding pending tasks to slave endpoints
>   - send pending tasks during reregistration of slave with master
> Isolation:
>   - replacing freezer with PID namespaces to avoid lost tasks during
> container destruction
>   - /tmp isolation
> Reliability:
>   - handling race between znode removal and state read
>   - allowing increase in available resources without requiring slave drain
>
> - dominic
>



-- 
Dominic Hamon | @mrdo | Twitter
*There are no bad ideas; only good ideas that go horribly wrong.*

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message