[jira] [Updated] (TS-3853) Request loop when reverse_proxy disabled

["Bryan Call (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/TS-3853?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Bryan Call updated TS-3853:
---------------------------
    Summary: Request loop when reverse_proxy disabled  (was: request loop when reverse_proxy disabled)

> Request loop when reverse_proxy disabled
> ----------------------------------------
>
>                 Key: TS-3853
>                 URL: https://issues.apache.org/jira/browse/TS-3853
>             Project: Traffic Server
>          Issue Type: Bug
>          Components: HTTP
>    Affects Versions: 5.3.1
>         Environment: Centos7
>            Reporter: Angelo Turetta
>             Fix For: 6.1.0
>
>
> I have 4 servers with TS 5.3.1 on Centos7 (self-built package). They are used as forward proxies by a lot of web servers that need to access external resources.
> Yesterday I applied this change to my records.config
> 151c151
> < CONFIG proxy.config.reverse_proxy.enabled INT 1
> ---
> > CONFIG proxy.config.reverse_proxy.enabled INT 0
> tonight a security scan took all the proxies to their knees with a simple HTTP GET on the service port.
> I reproduced the issue this morning: if the proxy IP address is 22.33.44.55, serving requests on port 8080, trying to 'wget -SO/dev/null http://22.33.44.55:8080/' when the reverse proxy is disabled, will trigger an infinite request loop that generates an Alert: 'too many connections, throttlin' and stops only when i restart the service.
> Of course I restored the reverse_proxy to enabled...
> is there some safeguard against this behavior?



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)