Hey Guys,
Thanks so much for the ideas and links. I've been doing some more thinking
around this, and was wondering if fashioning a single-threaded pull queue
would serve my purposes?
Basically, when a request comes in, push it to the pull queue, do the
linear assignment of the value through a
> Everything is managed for you. That means ANYTHING that relies on the
> 'machine' can fail randomly (and does)
this implication is totally wrong.
On Fri, Jul 26, 2013 at 3:27 PM, Richard wrote:
> One other item.
>
> I have had DeadlineExceeded Errors when calling time.time() in Python.
>
> Everything is managed for you. That means ANYTHING that relies on the
> 'machine' can fail randomly (and does)
http://www.zdnet.com/google-at-scale-everything-breaks-3040093061/
>
--
You received this message because you are subscribed to the Google Groups
"Google App Engine" group.
T
One other item.
I have had DeadlineExceeded Errors when calling time.time() in Python.
Everything is managed for you. That means ANYTHING that relies on the
'machine' can fail randomly (and does)
--
You received this message because you are subscribed to the Google Groups
"Google App Eng
Another idea to speed up synchronization on a high concurrency
transactional level, would be to create a single java backend instance that
takes in tasks from the front ends and process it in a singular queue. That
way you know the time is just from one server and you also know its from a
single th
If accuracy is that important to you, using URLFetch would not be accurate
enough.
If you use URLFetch, and the request takes say 400ms, you've no idea where
the actual timestamp belongs in that window. In theory its the middle but
because you don't know the relative delays of the send/receive pha
Hey guys,
I'm curious to hear about any strategies people have used to employ time
synchronization among the AppEngine instances. I'm working on a project
that is looking to have a pretty important time component and the current
strategy I'm looking at is something long the lines of:
1. Record
