Hi all, 

In top of our ceph cluster, one application use the rados gateway/S3. 
This application did not use multipart s3 api, but it split files (for example 
1 MB) into chunk of desire size (it have to work on top of several type of 
storage). 
For every tests, the application hang when uploading the 64th chunk ... And the 
upload failed. 
The chunk from 0 to 63 is well upload. 

For example, for a file of 100MB, the application would send 100 chunks of 1 
MB. But it hang when uploading the chunk 64 ... And i don't know why ... 
I did not find setttings of civetweb, where can we specify parameters for 
civetweb ? 
For example a setting which limit for 64 upload at the same time for the same 
client .... ? 

s3cmd is able to send the same file with the multipart s3 api, and setting 
chunk (5MB). But in our case, we do not use this api ... 

Thank you. 
Stephane. 

Here is the log 
2016-12-20 10:02:10.313809 7f00c52d9700 1 ====== starting new request 
req=0x7f00c52d3710 ===== 
2016-12-20 10:02:26.702572 7f00c52d9700 1 ====== req done req=0x7f00c52d3710 op 
status=0 http_status=200 ====== 
2016-12-20 10:02:26.702650 7f00c52d9700 1 civetweb: 0x7f01480743e0: 
100.74.189.131 - - [20/Dec/2016:10:02:10 +0100] "PUT 
/test-bucket/Gros%20fichiers/10G.63_96 HTTP/1.1" 200 0 - - 
2016-12-20 10:23:43.112493 7f00b1ab2700 1 ====== starting new request 
req=0x7f00b1aac710 ===== 
2016-12-20 10:23:43.121837 7f00b1ab2700 1 ====== req done req=0x7f00b1aac710 op 
status=-2010 http_status=400 ====== 
2016-12-20 10:23:43.121920 7f00b1ab2700 1 civetweb: 0x7f0138050490: 
100.74.189.131 - - [20/Dec/2016:10:23:43 +0100] "PUT 
/test-bucket/Gros%20fichiers/10G.64_96 HTTP/1.1" 400 0 - - 
2016-12-20 10:33:27.564398 7f00bc2c7700 1 ====== starting new request 
req=0x7f00bc2c1710 ===== 
2016-12-20 10:33:27.570324 7f00bc2c7700 1 ====== req done req=0x7f00bc2c1710 op 
status=0 http_status=200 ====== 
2016-12-20 10:33:27.570394 7f00bc2c7700 1 civetweb: 0x7f01752f8990: 
100.74.189.131 - - [20/Dec/2016:10:33:27 +0100] "GET / HTTP/1.1" 200 0 


_______________________________________________
ceph-users mailing list
ceph-users@lists.ceph.com
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com

Reply via email to