You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@trafficserver.apache.org by "Adam W. Dace (JIRA)" <ji...@apache.org> on 2015/07/11 06:28:04 UTC

[jira] [Comment Edited] (TS-3750) HostDB: Setting DNS TTL to "ignore" still creates many DNS lookups

    [ https://issues.apache.org/jira/browse/TS-3750?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14623195#comment-14623195 ] 

Adam W. Dace edited comment on TS-3750 at 7/11/15 4:27 AM:
-----------------------------------------------------------

Kinda bored, so I'm attempting to dig into this further.

At first I thought it was only HTTP proxy requests.  That assumption was just plain wrong.

Along the way I discovered I'm getting a heck of a lot of "DNS Error Code 0" errors.
Although I'm sure this is a symptom, not a cause, I've come up with a one-line patch
that allows HostDB to function more effectively.

I'll probably take another look at this in a few days.



was (Author: adace):
Kinda bored, so I'm attempting to dig into this further.

For some reason, this only occurs during HTTP proxy requests.  From what I can tell, HTTPS tunneling works fine...every time.
Still no idea why.

Back to digging.  :)

> HostDB:  Setting DNS TTL to "ignore" still creates many DNS lookups
> -------------------------------------------------------------------
>
>                 Key: TS-3750
>                 URL: https://issues.apache.org/jira/browse/TS-3750
>             Project: Traffic Server
>          Issue Type: Bug
>          Components: DNS, HostDB
>    Affects Versions: 6.0.0
>            Reporter: Adam W. Dace
>            Priority: Minor
>             Fix For: 6.1.0
>
>
> In an effort to further speed my forward-only proxy cache, I've been experimenting with using a 1-4 hour internal TTL in HostDB(as opposed to the default).  Many websites set their DNS TTL extremely low(i.e. 60 seconds), and I'd like to bypass that behavior.
> Unfortunately, according to OpenDNS my proxy is still generating roughly 1 DNS lookup per HTTP request.  I don't know why.
> I've taken a look at the codebase, but frankly it's all state machines in there and I don't possess the skills to identify and fix this problem.  I have turned on debugging and can say there does appear to be a lot of DNS activity going on.
> If someone could take a look, I'd greatly appreciate it.  I'll be happy to help test, also.  Thanks!



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