[Performance] Increased TTFB on assets


on one of my sites (https://rødt.no) I’m experiencing increased latency (TTFB) on assets, up to almost 4 seconds on some. That doesn’t seem right!


I’m not seeing this issue on other pages, so that adds to my confusion. Any clues as to what might be wrong?

Hi @jorngeorg, can you provide the URL for that specific asset that you took the screen shot from? Also in the response to that request there is an x-nf-request-id that allows us to lookup a request in our logs, that would also be useful to debug this issue. Thanks.

Hi thanks for getting back to me. Here’s the information you requested:

Url: https://rødt.no/images/logo_pos_anim.svg
x-nf-request-id: 8e454da7-12cf-4718-b020-1632d2d3383d-255093

So, I see that asset load at 2019-06-29 10:30:55 UTC and it seemed to take us 2 milliseconds to serve it, based on our server logs. Do you get that result reliably on that file? Does it happen to every asset or just that one?

I wasn’t able to get a long load myself, but I only tried from a few locations (EU/AU/US).

Hi, thanks for looking into this.

I’m getting fairly long TTFB on most of the local assets (images, css, js), but not on external resources. Some examples below:

Strangely I’m not getting long loads if I try to request the asset directly (copy/paste url):

That might be a hint that there’s something amiss in my front-end code, perhaps?

Sorry for the slow response, Jørn! Would you mind capturing a HAR file (see https://toolbox.googleapps.com/apps/har_analyzer/ for details about how to get one from Chrome) of those slow loads so we can dig in a bit more? The screenshot suggests that it is not what you mention (which was also what I was hoping the answer was!), but it’s hard to be sure without seeing a HAR file of you experiencing it.

Here’s the HAR-file: https://www.dropbox.com/s/goko9ujv3b7yqut/xn--rdt-0na.no.har?dl=1

(Hope I did that right)

Hi @jorngeorg,

I noticed quite a bit of queuing and stalled requests, however that may be separate from the issue you described which is a request that took you over 3 seconds, but from our side it completed much faster. Have you tested this on other networks to see if you have the same experience with long requests (not including stalled/queued requests)? Have you tried on Firefox? Note that I expect you to continue seeing the queuing/stalled requests, but taking multiple seconds to transfer a small file while on a fast connection is unusual.

I have also been seeing increased load times for assets, and it seems like TTFB is what’s doing it in my case, as well. As my site is completely static, I tried hosting it through good-old FTP and everything worked smoothly. When going back to Netlify, loading got slow again.

Requesting assets directly is quick/normal with me too.