You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@trafficserver.apache.org by Eric Duesterhaus <ed...@knapheide.com> on 2015/11/12 16:42:45 UTC
Troubleshooting a strange issue with missing scheme, and errored
requests.
Hi Again Everyone!
I'm having a strange issue that I'm not quite sure how to troubleshoot. About 2% of my requests are failing.
Each time a request fails it increments the count in traffic_logstats under URL Schemes for "none" or "other"
URL Schemes Count Percent Bytes Percent
------------------------------------------------------------------------------
HTTP (port 80) 7 0.01% 4.22KB 0.00%
HTTPS (port 443) 121,831 98.05% 1.23GB 100.00%
none 1,102 0.89% 0.00KB 0.00%
other 1,316 1.06% 0.00KB 0.00%
Also in traffic_logstats, it shows something... odd:
Origin Server Hits Misses Errors
------------------------------------------------------------------------------
originserver.example.com 0 121,854 0
dnsnamethatpointstoATS.example.com 0 7 1,316
==============================================================================
I have the following rules in remap.config:
#redirect to HTTPS
redirect http://dnsnamethatpointstoATS.example.com https://dnsnamethatpointstoATS.example.com
#remap rules to origin server
map https://dnsnamethatpointstoATS.example.com/SomeApplication https://originserver.example.com/SomeApplication
reverse_map https://originserver.example.com/SomeApplication https://dnsnamethatpointstoATS.example.com/SomeApplication
map https://dnsnamethatpointstoATS.example.com/ https://originserver.example.com/
reverse_map https://originserver.example.com/ https://dnsnamethatpointstoATS.example.com/
The number of errors at the beginning of the logstats output correlates to the count of "other" errors with missing URL schema. I'm not sure why some requests would be using "dnsnamethatpointstoATS.example.com" as the origin server, let alone with a Scheme of "other".
I get two types of errors in squid.blog when this happens:
1447283397.297 86400734 <ip address> ERROR_UNKNOWN(90)/000 0 GET dnsnamethatpointstoATS.example.com/SomeApplication/stuff - EMPTY/- -
1447283401.484 86400288 <ip address> ERROR_UNKNOWN(90)/000 0 - / - EMPTY/- -
Requests that appear to be successful look like this in the squid.blog:
1447283475.486 96 <ip address> TCP_MISS/200 1178 GET https://originserver.example.com/SomeApplication/stuff - DIRECT/originserver.example.com application/json
I can't help but think I've done something incredibly wrong or dumb in my remap.config.
Has anyone seen anything like this before? I'm not quite sure what to do next for troubleshooting, as I'm super-green to ATS. Anyone have any ideas or suggestions on what I should look at next, or focus my efforts on investigating?
Eric Duesterhaus
Re: Troubleshooting a strange issue with missing scheme, and errored requests.
Posted by Bryan Call <bc...@apache.org>.
It is most likely related to this issue:
https://issues.apache.org/jira/browse/TS-1935
I had a hard time reproducing it in June and haven't had time since to look
at it.
-Bryan
On Thu, Nov 12, 2015 at 7:42 AM, Eric Duesterhaus <
eduesterhaus@knapheide.com> wrote:
> Hi Again Everyone!
>
>
>
> I’m having a strange issue that I’m not quite sure how to troubleshoot.
> About 2% of my requests are failing.
>
>
>
> Each time a request fails it increments the count in traffic_logstats
> under URL Schemes for “none” or “other”
>
> URL Schemes Count Percent Bytes
> Percent
>
>
> ------------------------------------------------------------------------------
>
> HTTP (port 80) 7 0.01% 4.22KB
> 0.00%
>
> HTTPS (port 443) 121,831 98.05% 1.23GB
> 100.00%
>
> none *1,102* 0.89%
> 0.00KB 0.00%
>
> other *1,316* 1.06%
> 0.00KB 0.00%
>
>
>
> Also in traffic_logstats, it shows something… odd:
>
>
>
> Origin Server Hits Misses
> Errors
>
>
> ------------------------------------------------------------------------------
>
> originserver.example.com 0
> 121,854 0
>
> dnsnamethatpointstoATS.example.com 0 7
> *1,316*
>
>
> ==============================================================================
>
>
>
> I have the following rules in remap.config:
>
> #redirect to HTTPS
>
> redirect http://dnsnamethatpointstoATS.example.com
> https://dnsnamethatpointstoATS.example.com
>
> #remap rules to origin server
>
> map https://dnsnamethatpointstoATS.example.com/SomeApplication
> https://originserver.example.com/SomeApplication
>
> reverse_map https://originserver.example.com/SomeApplication
> https://dnsnamethatpointstoATS.example.com/SomeApplication
>
> map https://dnsnamethatpointstoATS.example.com/
> https://originserver.example.com/
>
> reverse_map https://originserver.example.com/
> https://dnsnamethatpointstoATS.example.com/
>
>
>
>
>
> The number of errors at the beginning of the logstats output correlates to
> the count of “other” errors with missing URL schema. I’m not sure why some
> requests would be using “dnsnamethatpointstoATS.example.com” as the
> origin server, let alone with a Scheme of “other”.
>
>
>
>
>
> I get two types of errors in squid.blog when this happens:
>
>
>
> 1447283397.297 86400734 <ip address> ERROR_UNKNOWN(90)/000 0 GET
> dnsnamethatpointstoATS.example.com/SomeApplication/stuff - EMPTY/- -
>
> 1447283401.484 86400288 <ip address> ERROR_UNKNOWN(90)/000 0 - / - EMPTY/-
> -
>
>
>
> Requests that appear to be successful look like this in the squid.blog:
>
> 1447283475.486 96 <ip address> TCP_MISS/200 1178 GET
> https://originserver.example.com/SomeApplication/stuff - DIRECT/
> originserver.example.com application/json
>
>
>
> I can’t help but think I’ve done something incredibly wrong or dumb in my
> remap.config.
>
>
>
> Has anyone seen anything like this before? I’m not quite sure what to do
> next for troubleshooting, as I’m super-green to ATS. Anyone have any ideas
> or suggestions on what I should look at next, or focus my efforts on
> investigating?
>
>
>
>
>
> Eric Duesterhaus
>