trafficserver-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Bryan Call <bc...@apache.org>
Subject Re: Question regarding disk write
Date Thu, 12 Sep 2019 16:13:50 GMT
Have you tried to log the via header with option 3 turn on (proxy.config.http.insert_response_via_str)?
 Also, there is a cache write logging field (I have been using this lately for another project)
cwr (https://docs.trafficserver.apache.org/en/8.0.x/admin-guide/logging/formatting.en.html?highlight=logging#admin-logging-fields-cache
<https://docs.trafficserver.apache.org/en/8.0.x/admin-guide/logging/formatting.en.html?highlight=logging#admin-logging-fields-cache>).

If you want to get some detailed information about if a URL can be cached you can use this
tool I wrote.  I have used to fix out cache hit ratios (we went from 90% to close to 100%):
https://github.com/apache/trafficserver/blob/master/tools/traffic_cacheable.pl <https://github.com/apache/trafficserver/blob/master/tools/traffic_cacheable.pl>

Example output:
[bcall@xxx ~]$ ./traffic_cacheable.pl cdn_url
Total Tested: 10
Response Codes:
	200: 10 (100.00%)
Freshness:
	0: 0 (0.00%)
	1 minute: 0 (0.00%)
	1 hour: 0 (0.00%)
	1 day: 1 (10.00%)
	1 week: 0 (0.00%)
	1 month: 5 (50.00%)
	1 year: 0 (0.00%)
	> 1 year: 1 (10.00%)
	undef: 3 (30.00%)
Freshness with Heuristic using Last-Modified (RFC 7234 4.2.2):
	0: 0 (0.00%)
	1 minute: 0 (0.00%)
	1 hour: 0 (0.00%)
	1 day: 4 (40.00%)
	1 week: 0 (0.00%)
	1 month: 5 (50.00%)
	1 year: 0 (0.00%)
	> 1 year: 1 (10.00%)
	undef: 0 (0.00%)
Headers:
	Cache-Control:
		max-age: 9 (90.00%)
		undef: 1 (10.00%)
	Expires:
		undef: 10 (100.00%)
	Last-Modified:
		defined: 10 (100.00%)
	Pragma:
		undef: 10 (100.00%)
Extra:
	Cache-Control / Expires not set: 1 (10.00%)
	Cache-Control / Expires / Last-Modified not set: 0 (0.00%)
	ATS Cache Hit: 2 (20.00%)

-Bryan


> On Sep 9, 2019, at 8:54 AM, Jason Yang <peter.waynechina@gmail.com> wrote:
> 
> Hi community, 
>      I am running an ATS cluster with debug workload (a replay of normal workload with
all headers removed) and I have noticed that a significant portion (>20%) of cache misses
are just tunneled through without being written to disk (extracted using via header, and in
my toy experiment, every miss should be cached). I assume this happens because the aggregation
write buffer is full, so the some of the disk writes are dropped. 
> 
> I have tried to increase the the aggregation buffer from 4MB to 256MB when I compile
ATS, but it does not help. 
> 
> Besides my NVME disk throughput (> 1600MB/s for 1MB fragment 80%read+20%write from
FIO test) is way higher than what I observed during the experiments, during which I rarely
see over 200MB/s read+writing. 
> 
> Any suggestion on how I should tune ATS to gain better performance? Maybe increase the
number of cache threads per disk? Or maybe increase/reduce fragment size? Or any suggestion
on how to figure out the reason of dropped disk writes? 
> Any thoughts are appreciated! 
> 
> 
> Best, 
> Jason 
> 
> 


Mime
View raw message