hi Joachim and all,
I am not aware of further plans concerning the specification of server
limits in the WADL.
Like you wrote, the limits are node-specific due to different
implementations or capabilities. Indeed, I wonder what would be the
relevant unit (acceptable and measurable at every node) for expressing
the service limits. This is maybe the reason why the FDSN spec is
(deliberately?) not accurate.
About the client being able to predict if he's going to hit with a 413
(eg. for fdsn-dataselect), I see an issue due to the presence of
wildcards in request parameters (not knowing in advance how wildcards
will expand at server level). One possibility would be to use the
channel inventories as provided by fdsnws-station, to compute and
predict (?) the data volumes.. (or even smarter, use the data
availability feature as proposed in fdsn-station).
best,
Pierre
RESIF node
On 02/22/2016 02:31 PM, Joachim Saul wrote:
Hello Working Group III,
fdsnws-dataselect sometimes responds with HTTP status 413 (Request
Entity Too Large). This requires reducing the request size by splitting
bulk requests into smaller pieces (fewer lines, possibly shorter time
windows) and submitting them separately.
It is clear that some limits are required in order to ensure operability
of the services. It is also clear that due to different implementations
and server capabilities at every fdsnws node the limits will have to be
node specific.
What is annoying from the user perspective is lack of a generic
mechanism to determine the server limits before submitting a request and
getting the 413 error. The returned error message text may contain a
description of the server limits but this is not mandatory.
The fdsnws specification 1.1 includes in table 1 with respect to error
413 the recommendation "Service limits should also be documented in the
service WADL." Even though this sounds like it could solve the problem
of unknown request limits, there is AFAICS no "official"
fdsnws-dataselect server where this feature has already been implemented.
Considering the usefulness especially in the context of federated data
centers, shouldn't the specification of server limits in the WADL be
made mandatory? Are there any plans in that direction?
Regards
Joachim
----------------------
FDSN Working Group III (http://www.fdsn.org/message-center/topic/fdsn-wg3-products/)
Sent via IRIS Message Center (http://www.fdsn.org/message-center/)
Update subscription preferences athttp://www.fdsn.org/account/profile/
Joachim Saul
Pierre Volcke