Search Mailing List Archives


Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[bioontology-support] [ncbo-annotator-users] 500 errors from the Annotator service

Ray Fergerson ray.fergerson at stanford.edu
Thu Jun 23 12:36:02 PDT 2011


You are correct that we would rather you didn't. Others have though and
this has caused problems. We don't currently have a clean way to throttle
user requests.

 

Ray

 

From: Benjamin Good [mailto:ben.mcgee.good at gmail.com] 
Sent: Thursday, June 23, 2011 12:29 PM
To: Ray Fergerson
Cc: support at bioontology.org; annotator-discuss at googlegroups.com; Taciana
Gomes
Subject: Re: [bioontology-support] [ncbo-annotator-users] 500 errors from
the Annotator service

 

Ray,

 

Correct, only one thread.  (I'd happily launch an armada for some jobs if
that was fine with you, but I assume that would not be well-received).

 

-Ben

On Thu, Jun 23, 2011 at 12:02 PM, Ray Fergerson
<ray.fergerson at stanford.edu> wrote:

Ben,

 

Thanks for the helpful information. Just confirming that you are making a
call and waiting for a return before you make another call. You don't have
1000 threads beating on the system at the same time. Correct?

 

Note, as a side matter unrelated to this instability, I believe that the
longestOnly parameter only works correctly if you keep all ontologies. (I
also thought though that we detected this combination of parameters now
and blocked the call. Obviously I am wrong about this latter.) This
limitation is really a design problem and is unlikely to get addressed
anytime soon. The "longest only" logic is implemented by a separate
process that knows nothing about ontologies. The filtering of results on
ontologies is done after this matching. I suggest turning off the longest
only parameter and doing the logic yourself on the returned values. 

 

Ray 

 

From: Benjamin Good [mailto:ben.mcgee.good at gmail.com] 
Sent: Thursday, June 23, 2011 11:32 AM
To: Ray Fergerson
Cc: support at bioontology.org; annotator-discuss at googlegroups.com; Taciana
Gomes


Subject: Re: [bioontology-support] [ncbo-annotator-users] 500 errors from
the Annotator service

 

Hi Ray,

 

I had failures last night and this morning.

 

The client source code that accesses the service can be found here:

http://code.google.com/p/genewiki/source/browse/java/Miner/src/org/gnf/ncb
o/web/AnnotatorClient.java

 

If you want me to make you an executable for testing let me know.  (That
code isn't really ready for primetime).. Since the problem was also
observed by some one running a c# client I hope its not a problem with my
implementation.

 

If I repeatedly run the main method in that class (or execute the request
from other resources that depend on it) sometimes it works and sometimes
it doesn't (responding with a 500).  I'm sending pretty small chunks of
text (typically less than a sentence) at irregular intervals.  If it
fails, I wait 2 seconds and try it again.  If it fails 10 times the client
quits.  That retry loop was sufficient to overcome similar behavior prior
to this update for tens of thousands of requests sent right after each
other.  

 

params:

params.put("isVirtualOntologyId", "true");

params.put("rqnum", "0");

params.put("textToAnnotate", text2annotate);

params.put("filterNumber", "true");

params.put("minTermSize", "3");

params.put("withSynonyms", "false");

params.put("longestOnly", "true");

params.put("wholeWordOnly", "true"); 

params.put("stopWords", "protein, gene");

params.put("withDefaultStopWords", "true");

params.put("scored", "true");

params.put("mappingTypes", "null");

ontologiesToKeepInResult (virtual ids for GO, DO, FMA, PRO, OMIM)

 

Let me know if thats enough to get you going.  If not I will try to code
something that consistently reproduces your error (difficult because, from
my end, there is significant randomness on the server side).

 

thanks

-Ben

On Thu, Jun 23, 2011 at 11:09 AM, Ray Fergerson
<ray.fergerson at stanford.edu> wrote:

Ben,

 

Do you have a reproducible test case for this? Certainly the entire system
behavior was erratic before last week for several weeks. We believe that
this problem has been fixed though. If you are still seeing failures then
we would like to look into it. It will be difficult to do so though
without a test showing failures. We haven't had other complaints that I am
aware of since we pushed out the patch last week.

 

Even if it is not a single call that fails, perhaps you have a script that
we could execute that would reliably show failures. Also it would be very
useful to know how many calls you are making in some unit of time. Also
relevant are the parameters you are passing to the annotator and the type
and size of text.

 

Ray

 

From: bioontology-support-bounces at lists.stanford.edu
[mailto:bioontology-support-bounces at lists.stanford.edu] On Behalf Of
Benjamin Good
Sent: Thursday, June 23, 2011 9:40 AM
To: support at bioontology.org
Cc: annotator-discuss at googlegroups.com; Taciana Gomes
Subject: Re: [bioontology-support] [ncbo-annotator-users] 500 errors from
the Annotator service

 

For the last three weeks or so, the Annotator has become very unreliable.
It is reporting HTTP 500 errors for a large proportion of requests.
Repeating identical requests sometimes succeeds and sometimes do not.

 

These problems appear to coincide with the introduction of the new
authentication policy.  

 

Please let us know what is going on.  I have a live service that depends
on the Annotator...

 

thanks

-Ben

 

On Fri, Jun 10, 2011 at 12:55 PM, Taciana Gomes
<taciana.alessandra at gmail.com> wrote:

I did make a client with c#

 

---------- Forwarded message ----------
From: Taciana Gomes <taciana.alessandra at gmail.com>
Date: 10 June 2011 16:46
Subject: Re: [ncbo-annotator-users] 500 errors from the Annotator service
To: ben.mcgee.good at gmail.com


Hi, I have this problem too.
I dont know what can be.. 

 

On 8 June 2011 18:06, Benjamin Good <ben.mcgee.good at gmail.com> wrote:

I've been experiencing a large number of HTTP 500 errors from the
Annotator service over the past few days.  

 

Repeating exactly the same request sometimes produces an error and
sometimes does not..  My client currently tries a request up to ten times
before quitting - sometimes the first one works, sometimes the third,
sometimes the 5th and sometimes never..

 

What going on?

 

thanks

-Ben





-- 
____________________________________________
Taciana Alessandra Gomes Cruz
MSc in Computer Science
Federal University of Pernambuco - UFPE/Brazil
Fone: (19) 8185-8516




-- 
____________________________________________
Taciana Alessandra Gomes Cruz
MSc in Computer Science
Federal University of Pernambuco - UFPE/Brazil
Fone: (19) 8185-8516

 

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.stanford.edu/pipermail/bioontology-support/attachments/20110623/a0aa0e2f/attachment.html>


More information about the bioontology-support mailing list