Hi Jeff.

Thanks for your reply. I did some measurements.

First is a request made to a different remote host:

LTA Fetch on App Engine
===================

#
04-06 04:35PM 03.011 /api/v1/16061/30/ 200 768ms 168ms-cpu 0kb Mozilla/
5.0 (X11; U; Linux i686; en-US; rv:1.9.1b3) Gecko/20090305 Firefox/
3.1b3,gzip(gfe)
See details 218.186.10.253 - - [06/Apr/2009:16:35:03 -0700] "GET /api/
v1/16061/30/ HTTP/1.1" 200 275 - "Mozilla/5.0 (X11; U; Linux i686; en-
US; rv:1.9.1b3) Gecko/20090305 Firefox/3.1b3,gzip(gfe)"

#
I 04-06 04:35PM 03.027
Method get called for APIEndPoint

#
D 04-06 04:35PM 03.036
Cache hit for stop 16061

#
I 04-06 04:35PM 03.043
Fetching URL:
http://www.publictransport.sg/public/ptp/en/Getting-Around/ArrivaltimeResults.html?hidServiceNoValue=&hidBusStopValue=16061

#
I 04-06 04:35PM 03.626
Fetched URL:
http://www.publictransport.sg/public/ptp/en/Getting-Around/ArrivaltimeResults.html?hidServiceNoValue=&hidBusStopValue=16061

=========

As you can see, this urlfetch took just over half a second to
complete.

Now a request that caused a fetch from sbstransit.com.sg

NextBus fetch on App Engine
======================
#
04-06 04:34PM 11.864 /api/v1/17069/030/ 200 3276ms 279ms-cpu 0kb
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.1b3) Gecko/20090305
Firefox/3.1b3,gzip(gfe)
See details 218.186.10.253 - - [06/Apr/2009:16:34:15 -0700] "GET /api/
v1/17069/030/ HTTP/1.1" 200 173 - "Mozilla/5.0 (X11; U; Linux i686; en-
US; rv:1.9.1b3) Gecko/20090305 Firefox/3.1b3,gzip(gfe)"

#
I 04-06 04:34PM 12.073
Method get called for APIEndPoint

#
I 04-06 04:34PM 12.107
Fetching URL: 
http://www.sbstransit.com.sg/mobileiris/index_svclist.aspx?stopcode=17069

#
W 04-06 04:34PM 15.120
      Error fetching url 
http://www.sbstransit.com.sg/mobileiris/index_svclist.aspx?stopcode=17069
      Traceback (most recent call last):
        File "/base/data/home/apps/sbsnextbus/6.332607848382532409/
utils.py", line 72, in get_url
          result = urlfetch.fetch(url)
        File "/base/python_lib/versions/1/google/appengine/api/
urlfetch.py", line 273, in fetch
          raise DownloadError(str(e))
      DownloadError: ApplicationError: 5

=======================

Now, the urlfetch got cut-off in 3 seconds. This is longer than usual
but still under the 5 second limit. Now for comparison, here is a log
of the same request executed on my localhost in the development
server.

On localhost
==========

INFO     2009-04-06 23:50:15,961 main.py] Method get called for
APIEndPoint
DEBUG    2009-04-06 23:50:15,962 nextbus.py] Cache hit for stop 17069
INFO     2009-04-06 23:50:15,962 utils.py] Fetching URL:
http://www.sbstransit.com.sg/mobileiris/index_mobresult.aspx?stop=17069&svc=051
DEBUG    2009-04-06 23:50:15,964 urlfetch_stub.py] Making HTTP
request: host = www.sbstransit.com.sg, url =
http://www.sbstransit.com.sg/mobileiris/index_mobresult.aspx?stop=17069&svc=051,
payload = None, headers = {'Host': 'www.sbstransit.com.sg', 'Accept':
'*/*'}
DEBUG    2009-04-06 23:50:16,077 urlfetch_stub.py] Making HTTP
request: host = www.sbstransit.com.sg, url =
http://www.sbstransit.com.sg/mobileiris/(yvyrjnatocvc0y55rjqlocu3)/index_mobresult.aspx?stop=17069&svc=051,
payload = None, headers = {'Host': 'www.sbstransit.com.sg', 'Accept':
'*/*'}
INFO     2009-04-06 23:50:17,259 utils.py] Fetched URL:
http://www.sbstransit.com.sg/mobileiris/index_mobresult.aspx?stop=17069&svc=051
DEBUG    2009-04-06 23:50:17,260 nextbus.py] Saved cache for stop, svc
17069, 051
INFO     2009-04-06 23:50:17,274 dev_appserver.py] "GET /api/
v1/17069/051/ HTTP/1.1" 200 -

=============

On localhost, the urlfetch completed in just under 1.5 seconds.
Doesn't this indicate that (a) this could be a latency issue between
Google's servers & SBS Transit's servers? and (b) request should still
not be cut-off since it is well under the ~5 seconds limit.

I know the 5 seconds limit is not documented (at least I can't find
it) but it is an issue if it keeps changing all the time.

Any thoughts?

Cheers,

Deepak


On Apr 7, 5:16 am, Jeff S <[email protected]> wrote:
> Hi Deepak,
>
> The error you are seeing usually means that the urlfetch service did
> not receive the reply within ~5 seconds. There could be several
> possible sourced of latency which could be involved as you noted, so
> do you think the issue could be that the sbstransit site is responding
> more slowly than it used to?
>
> Thank you,
>
> Jeff
>
> On Apr 6, 5:45 am, Deepak Sarda <[email protected]> wrote:
>
> > Hi
>
> > One of my apps is getting DownloadError (ApplicationError 5) for
> > pretty much all urlfetch remote url fetch requests to a particular
> > remote server for the past week. Fetch requests to other servers are
> > not affected. Also, there are no problems fetching content from this
> > remote server from the local development server. Here's a traceback
> > from the logs:
>
> > Error fetching 
> > urlhttp://www.sbstransit.com.sg/mobileiris/index_svclist.aspx?stopcode=1...
> > Traceback (most recent call last):
> >   File "/base/data/home/apps/sbsnextbus/6.332480362470588654/utils.py",
> > line 71, in get_url
> >     result = urlfetch.fetch(url)
> >   File "/base/python_lib/versions/1/google/appengine/api/urlfetch.py",
> > line 273, in fetch
> >     raise DownloadError(str(e))
> > DownloadError: ApplicationError: 5
>
> > As can be seen from the trace, my app (http://sbsnextbus.appspot.com)
> > is trying to fetch pages from sbstransit.com.sg
>
> > This appears to be a deadline exceeded error (error code 5) but I
> > can't figure out why this should occur. Everything was running
> > smoothly till last week when this started happening.
>
> > Since I can't reproduce this problem locally and there is no problem
> > fetching pages from other remote servers in the production
> > environment, I suspect the issue is related to network connectivity
> > issues between AppEngine data centers & the servers ofwww.sbstransit.com.sg
>
> > This problem has been around for almost a week now and I can't seem to
> > figure out a solution or work-around. Searches in the group archive
> > didn't yield anything helpful. Any help would be greatly appreciated!
>
> > Thanks,
>
> > Deepak
>
> > PS: Is there a process to open a 'support ticket' with appengine team
> > regarding infrastructure issues?
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"Google App Engine" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to 
[email protected]
For more options, visit this group at 
http://groups.google.com/group/google-appengine?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to