You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@trafficserver.apache.org by "Bryan Call (JIRA)" <ji...@apache.org> on 2016/08/16 17:45:20 UTC
[jira] [Updated] (TS-2942) Weird value on % field in log
files
[ https://issues.apache.org/jira/browse/TS-2942?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Bryan Call updated TS-2942:
---------------------------
Summary: Weird value on %<fsiz> field in log files (was: weird value on %<fsiz> field in log files)
> Weird value on %<fsiz> field in log files
> -----------------------------------------
>
> Key: TS-2942
> URL: https://issues.apache.org/jira/browse/TS-2942
> Project: Traffic Server
> Issue Type: Bug
> Components: Logging, Metrics
> Reporter: Daniel Picolli Biazus
> Fix For: sometime
>
>
> We've configured ATS as a reverse proxy, and We've been noticed a really weird traffic value in our monitoring system. After spend some time digging through log files, I could notice that, in some cases the field %<fsiz> brings a huge value (20 characters) instead of the file size.
> I found this closed issue regarding the <fsiz> implementation, and I think there might be a issue when there is no Content-Length from the origin server.
> https://issues.apache.org/jira/browse/TS-2212
> Log format:
> ##########################
> <LogFormat>
> <Name = "myformat"/>
> <Format = "%<ttmsf> %<{Host}cqh> %<chi> [%<cqtn>] %<cqhm> %<cquup> %<cqhv> %<pssc> %<cqhm> %<fsiz> %<cquup> %<{User-Agent}cqh> %<{Referer}cqh> %<crc>"/>
> </LogFormat>
> ##########################
> Log result:
> ##########################
> 0.181 myhostname.com 123.123.123.123 [17/Jul/2014:15:50:33 -0000] GET /foo/bar/file.jpg HTTP/1.1 200 GET 3904675161847313968 /foo/bar/file.jpg Mozilla/5.0 (Windows NT 6.2; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/35.0.1916.153 Safari/537.36 http://myreferer.com TCP_MISS
> ###########################
> In that case, the <fsiz> field has the value "3904675161847313968" instead the value of a few bytes.
> I believe when there is no way to get the real content-length, we should return 0 in order to avoid misinterpretation.
> Thanks in advance,
> Daniel
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)