Recent activity
Subscribe to this feed
Olivier replied on September 23, 2009 20:29 to the question "Frequent timeouts when "checking for bandwidths"" in Connected Flow:
As an illustration, here are some graphs over 18 hours of the time it takes to do getUploadStatus and getInfo on my account. Probing interval was 15 minutes. It looks like I can keep getUploadStatus' time below 120 seconds if I probe every 5 minutes, which is interesting (and means that a workaround for me would be to have a cron keeping Flickr cache warm).

Olivier replied on September 22, 2009 21:18 to the question "Frequent timeouts when "checking for bandwidths"" in Connected Flow:
Olivier replied on September 22, 2009 15:33 to the question "Frequent timeouts when "checking for bandwidths"" in Connected Flow:
Having a look at my early numbers, a 120-second timeout would be a significant improvement but the safest value is more likely to be around 160 seconds. flickr.people.getInfo always returns within 5 seconds, though, so calling flickr.people.getUploadStatus only for non-pro accounts would be a definite win for all big users in my opinion.
Olivier replied on September 21, 2009 22:36 to the question "Frequent timeouts when "checking for bandwidths"" in Connected Flow:
Alright, I have a script ready. I'll gather some statistics and let you know. Looking at the Flickr API, though, I was wondering if it would be possible for you to first call flickr.people.getInfo, see if ispro is 1 -- in which case there's no bandwidth limit -- and then call flickr.people.getUploadStatus only if it's relevant to know the limit?
Not sure it would be faster (I'll hopefully figure out soon)... Just a thought.
Olivier replied on September 21, 2009 07:45 to the question "Frequent timeouts when "checking for bandwidths"" in Connected Flow:
Olivier asked a question in Connected Flow on September 20, 2009 10:26:
Frequent timeouts when "checking for bandwidths"Hi,
I've been having a lot of timeouts when "checking for bandwidth limits" over the last few months. I have a lot of pictures (close to 20k) and sets (several hundreds). It looks like it ends up working if I try several times in a row, which may mean that FlickrExport just needs to hit the same Flickr frontend twice and benefit from a cache at some level. According to tcpdump, it happens during flickr.people.getUploadStatus. Tcpdumps of typical failing and succeeding exchanges available upon request (just let me know if I need to obfuscate more than just the keys).
I was wondering if there's a way to tweak FlickrExport's timeout for my own needs or, since I have a Pro account with unlimited bandwidth, if I can just disable the bandwidth limit check?
Thanks.
Loading Profile...
