Are the response times slow on the ATS server?  If so, you can turn on slow log and see where the requests are slowing down.  Can you give some details on the type of requests TLS/SSL vs non-encrypted or http/1.1 vs http/2?

Documentation on slow log:

Example for slow log (times in seconds):
$ sudo /usr/local/bin/traffic_ctl config set proxy.config.http.slow.log.threshold 1
$ tail /usr/local/var/log/trafficserver/diags.log
[Oct 19 08:27:20.578] Server {0x2b7839272700} ERROR: [2] Slow Request: client_ip: url: status: 301 unique id:  redirection_tries: 0 bytes: 304 fd: 0 client state: 0 server state: 9 ua_begin: 0.000 ua_first_read: 0.000 ua_read_header_done: 0.000 cache_open_read_begin: 0.000 cache_open_read_end: 0.000 dns_lookup_begin: 0.000 dns_lookup_end: 0.000 server_connect: 0.000 server_first_read: 0.241 server_read_header_done: 0.241 server_close: 0.241 ua_close: 0.241 sm_finish: 0.241 plugin_active: -0.001 plugin_total: -0.001


On Oct 19, 2016, at 4:46 AM, Paolo Barbato <> wrote:

I’m using ATS as forward proxy, with a cache of 50 GB on disk, and 100/150 clients. In front there’s a CheckPoint  firewall.

ATS is installed on a CentOS 7.2 with 9 GB ram hosted on a VMWare farm.

It works mostly fine, but time to time users (and me) observe hang when loading pages, not specific urls.

CPU load is really low as well as memory usage (no swap at all).

Monitoring cache port 8080, I see more or less 600/800 established connection per seconds during peak hours, with growing closed one that top at 500 at end of the day .

If clients bypass ATS no issues are observed.

Any hints on possible reasons?


Paolo Barbato

Consorzio RFX
corso Stati Uniti,4                                  
35127 Padova - Italy                                       
Network Administrator 
phone: +39 049 8295097 fax: +39 049 8700718