Initial seeks can be overly slow

From @vtsukur on Tue Sep 29 2015 18:07:11 GMT+0000 (UTC)

From time to time it can be observed that initial seek for a file can be overly slow.

For example, when file is ipfs add-ed on a locally running machine, and then retrieved via gateway.ipfs.io/ipfs, request can take any time in between several hundreds of milliseconds down to 2 minutes timing out several times in a row (ending in successful seek, eventually).

If file is ipfs get-ed :slight_smile: from a node that does not share same network, it is usually much much slower. Succeeds, but amount of attempts is usually bigger.

I am curious to know why these latency issues are present at the moment. Thanks!


Copied from original issue: https://github.com/ipfs/faq/issues/57

From @jbenet on Wed Sep 30 2015 00:45:36 GMT+0000 (UTC)

this sounds like NAT problems and nodes not finding you. these will get a lot better, and in the general case they already are.

From @jbshirk on Thu Oct 29 2015 14:20:10 GMT+0000 (UTC)

I have observed the same, but assumed this was just normal propagation time. But if itโ€™s NAT problems, where should people go to learn how to resolve that?

From @RichardLitt on Mon May 02 2016 19:24:39 GMT+0000 (UTC)

@whyrusleeping I believe some of this was fixed in 0.4.0 and 0.4.1. Can you confirm?

I just installed version 0.4.10, and I had the same problem. I followed the Getting Started instructions here https://ipfs.io/docs/getting-started/, and it hung on the curl GET for minutes. A retry did work though.

Also, my node was apparently very popular for other pplโ€™s content because it filled up my network :grin: