For the first time I've had to wrap something I'm working on as a CGI script. I'm having trouble with browsers (Both both Chrome and Firefox) not recognising the Content-Length header and stating size "unknown" to the users.
When I test this with the linux too wget
, the tool recognises the size just fine.
When I test this manually though openssl s_client -connect
I get the following headers:
The precise output from the webserver is as follows:
HTTP/1.1 200 OK
Date: Sun, 30 Jul 2017 20:12:20 GMT
Server: Apache/2.4.25 (Ubuntu) mod_fcgid/2.3.9 OpenSSL/1.0.2g
Content-Disposition: attachment; filename=foo.000000000G-000000001G.foofile.txt;
Content-Length: 501959790
Vary: Accept-Encoding
Content-Type: text/plain;charset=utf-8
Can anyone suggest what is missing / badly formatted?
Cracked it eventually.
This was caused by Apache doing something unexpected. Apache is compressing the output of the CGI script on the fly (sending with Content-Encoding: gzip
). This changes the size of the file but Apache cannot know how much it is going to change when it sends the header. The files are 1/GB each so it can't / doesn't cache the gzipped content before it starts sending therefore cannot know the file size. This means it has to switch to Transfer-Encoding: chunked
One way to fix this is set Content-Encoding: none
in the header which disables Apache from compressing the content. This does mean that 1/2 GB files take much longer to send.
Another might be to manually gzip the content in my cgi script and setting Content-Encoding: gzip
and Content-Length: <gzipped size>
. This will require me to work out the compressed size before sending.
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.