Thank you for popup this, have already mark it resolved.
On Mon, May 2, 2016 at 4:32 AM, Marco Massenzio
wrote:
> Hi,
>
> sorry, I have not kept up with all the new endpoints :)
> If there is already an endpoint (/redirect ?) that essentially addresses
> the issue raised
Hi,
sorry, I have not kept up with all the new endpoints :)
If there is already an endpoint (/redirect ?) that essentially addresses
the issue raised by MESOS-3841
(https://issues.apache.org/jira/browse/MESOS-3841) then I'd suggest to
close it and add a note.
(I just saw it and thought it would
Oh, @Marco. Thank you very much for your reply, vinodkone shepherd this and
it have already submitted after other kindly guys reviews.
For MESOS-3841, it should be resolved now because could get the leading
master by "/redirect" endpoint. Do you have any concerns about it? I would
like to solve
@haosdent - thanks for doing this, very useful indeed!
On a related issue [0], I'd like to that one on:
- can anyone comment if that's a good idea/bad idea; and
- would anyone be willing to shepherd it?
Thanks!
[0] Master HTTP API support to get the leader (
Hi All,
We intend to introduce a breaking change[1] in the http endpoints without
the deprecation cycle.
For below http endpoints, when user request to a master which is not a
leader,
user would get a 307 redirect(TEMPORARY_REDIRECT) to the leader master.
* /create-volumes
*
Some feedback on this ticket: it focuses on the solution rather than the
problem. We generally want to avoid this, I guess it's been coined 'The XY
Problem' (thanks Benjamin Bannier). In this case it turns out that there
are actually 2 distinct problems that the user is facing:
(1) Passive
+1
(my two cent is that the “correct” approach from an operations viewpoint is to
first query for the leader, then ask the leader; shortcoming identified by Ben
obvious, but possibly the lesser of the two evils - and probably unavoidable in
a distributed systems without atomic transactions -
On Fri, Jan 8, 2016 at 12:29 PM, Benjamin Mahler wrote:
> (2) It is difficult to reliably obtain cluster state through the existing
> endpoints. This one is less clear to me than the first problem. Here we
> have to think through how we want users to be hitting state
We should add the "who-is-the-current" leader informational endpoint
regardless of whether we do redirection, no?
Will it be clear which endpoints should redirect? Seems the redirection
approach, if we were to do it, needs to be specified explicitly by the
user. Otherwise it may be confusing for
Alright, let's revive it. I think we previously had problems trying to
write a multi-master unit test. Might have to do some test infrastructure
work to make that possible.
On Wed, Jan 6, 2016 at 1:14 PM, Neil Conway wrote:
> +1 -- I think we should make this change. The
Regarding this issue, I see non-active marathon instance would proxy http
requests to the active marathon instance. This way no matter which marathon
instance the client is visiting, it would always get the correct result.
Could we do the same with mesos masters? The implementation would be more
Hi, Adam and Haosdent
Resurrecting this issue, https://issues.apache.org/jira/browse/MESOS-1865, I
would like to make a +1 for this change, which apparently became cold but I
think is very relevant and we had enough time to be prepared for a change like
this, right?
If necessary, can I help
+1 -- I think we should make this change. The current behavior is
quite dangerous.
Neil
On Wed, Jan 6, 2016 at 12:52 PM, Diogo Gomes wrote:
> Hi, Adam and Haosdent
>
>
> Resurrecting this issue, https://issues.apache.org/jira/browse/MESOS-1865, I
> would like to make a +1
Hi Haosdent,
Do you have a shepherd for this change?
On 1 Jul 2015 8:06 am, Adam Bordelon a...@mesosphere.io wrote:
The original ticket MESOS-1865
https://issues.apache.org/jira/browse/MESOS-1865 argues the point that
requesting state data (e.g. tasks.json) from a non-leading master should
I'll volunteer to shepherd this, unless somebody else really wants to. I
think returning any HTTP error code (other than 200 OK) would be preferable
for those endpoints that would return empty data otherwise, but redirect is
the most actionable one for getting to the real data.
On Wed, Jul 1,
Sure, that makes sense. I guess I was wondering if we should document a
recommended retry-limit threshold for people writing clients - along with a
recommended approach for backing off before retrying again.
On Wed, Jul 1, 2015 at 12:29 PM, haosdent haosd...@gmail.com wrote:
Hi, @James Thank
Oh, I got it. I would discuss document this with @adam. Thank you for your
great advice.
On Thu, Jul 2, 2015 at 12:33 AM, James DeFelice james.defel...@gmail.com
wrote:
Sure, that makes sense. I guess I was wondering if we should document a
recommended retry-limit threshold for people writing
Hi, @Tomas Senart. Currently, my patch is to redirect the request to
correct url. For example:
$ curl -i http://master1:5050/master/tasks.json
HTTP/1.1 307 Temporary Redirect
Date: Mon, 01 Jun 2015 06:30:08 GMT
Location: http://master2:5050/master/tasks.json
Content-Length: 0
Assume master1 is
In a cluster that's having network problems and the selected leader is
flapping is there an upper limit on how many subsequent redirects a
client may expect to receive before it should give up for some period of
time? For example:
client requests /tasks.json from master1
master1 sends redirect to
Hi, @James Thank you very much for your good question. My current patch
could not avoid this problem. I think you could handle this in client side,
give it up or return error when redirect times overflow your define limit.
On Thu, Jul 2, 2015 at 12:23 AM, James DeFelice james.defel...@gmail.com
+1
As an API writer (and consumer), this behavior makes sense, and most
clients (notably, Apache HttpClient) would be able to handle this
transparently
(I think - it's been a few years since I messed with it :)
I completely agree that returning a 200 OK with empty (or, worse, stale)
data would
The original ticket MESOS-1865
https://issues.apache.org/jira/browse/MESOS-1865 argues the point that
requesting state data (e.g. tasks.json) from a non-leading master should
not return 200 OK status code with empty data, as that is misleading. It
should either return an error code, or valid data.
Hi All,
We intend to introduce a breaking change[1] in the http endpoints. For
below http endpoints, when user request to a master which is not a leader,
user would got a 302 redirect to the leader master.
* /slaves
* /state
* /stateSummary
* /roles
* /teardown
* /tasks
Hi Haosdent,
Thanks for the heads up. Would you be able to share the rationale for this
change? Is it a precursor for something else?
Best,
Tomás
On Tue 30 Jun 2015 at 19:24 haosdent haosd...@gmail.com wrote:
Hi All,
We intend to introduce a breaking change[1] in the http endpoints. For
24 matches
Mail list logo