Big files truncated

[Axel]

Hi Maxim, 

thanks for your reply. I don't think firewall could have anything to do with this. And I don't find any log related to crashed workers. 
All I have is: 
2010/07/28 18:19:49 [info] 28987#0: *49 client closed prematurely connection while sending response to client, client: 213.31.XX.XX
but I think this is not an issue as well.
Nothing in nginx general logs, nothing in syslog

Is there any limit (memory, timeout) I should increase?

Thanks

Axel


Le 17 août 2010 à 16:42, Maxim Dounin a écrit :

> Hello!
> 
> On Tue, Aug 17, 2010 at 03:00:17PM +0200, Axel wrote:
> 
>> I get still the same issue on nginx 0.7.67. Medium and big files 
>> are sometimes (I don't know when or why) truncated. I have a 
>> server configuration for static.mydomain.tld that is doing 
>> nothing else than sending static content (no PHP, CGI ...). 
>> 
>> server {
>>       listen 80;
>>       server_name static.mydomain.tld;
>>       root /var/www/medias/;
>>       expires 90d;
>>       access_log /var/log/nginx/medias_access.log;
>>       error_log  /var/log/nginx/medias_error.log notice;
>> 
>>       location /videos/ {
>>               expires -90d;
>>               send_timeout          600;
>>               keepalive_timeout     600 600;
>>               error_page 404 = /videos/video_not_found.png;
>>       }
>> 
>>       location /gallery/ {
>>               error_page 404 = /galerie/1x1.png;
>>       }
>> }
>> 
>> When it fails on images, I have half-downloaded images and when 
>> it fails on videos, the video playback stops after a couple a 
>> seconds.  Here is a screenshot of failed images
> 
> Config looks sane and shouldn't cause any problems.
> 
> Most likely reason is shortage of some system resources, e.g.  
> you've run out of memory or socket buffers.  Other possible 
> reasons include:
> 
> - something like statefull firewall inbetween which run 
> out of states and dropping random connections;
> 
> - nginx workers dying for some reason, likely unrelated to this 
> particular server block.
> 
> Looking into error_log may be helpfull.  Note well: you may want 
> to look into global error_log as well, not only the one defined 
> for this particular server.
> 
> Maxim Dounin
> 
> _______________________________________________
> nginx mailing list
> nginx at nginx.org
> http://nginx.org/mailman/listinfo/nginx