[squid-users] Squid returns a lot of ABORTED in access log and user navigation speed slows

[Jonathan Lee]

I think they are default on if you look at references 


> On May 15, 2024, at 17:16, Andre Bolinhas <andre.bolinhas at articatech.com> wrote:
> 
> Hi Alex
> Thanks for your feedback, in this case enable client_persistent_connections and server_persistent_connections could help or is better to keep it disable?
> 
> Best regards
> 
> On 15/05/2024 19:24, Alex Rousskov wrote:
>> On 2024-05-15 14:08, Andre Bolinhas wrote: 
>> 
>>> I'm not using pipeline_prefetch, because pipeline_prefetch breaks the NTLM/Kerberos authentication.
>> 
>> 
>> Enabling pipeline_prefetch introduces other problems as well. There might be some very special use cases that benefit from pipeline_prefetch today, but, in general, that directive should not be used (and the whole feature should be removed from Squid until it is properly implemented). 
>> 
>> I cannot currently answer your primary questions on this thread. I hope somebody else will guide you through this triage. 
>> 
>> Alex. 
>> 
>> 
>>> On 15/05/2024 18:15, Jonathan Lee wrote:
>>>> Have you researched enabling pipeline_prefetch?? 
>>>> 
>>>>> On May 14, 2024, at 17:56, Andre Bolinhas <andre.bolinhas at articatech.com> <mailto:andre.bolinhas at articatech.com> wrote: 
>>>>> 
>>>>> Hi 
>>>>> 
>>>>> Sometimes my users complains that the internet navigation thought Squid is very slow. 
>>>>> 
>>>>> After checking the access.log, I can see a lot of ABORTED messages like this 
>>>>> 
>>>>> 1715537802.589      2 10.103.12.94 NONE_NONE_ABORTED/200 0 CONNECT api.telegram.org:443 - HIER_NONE/-:- - mac="00:00:00:00:00:00" accessrule:%20global_whitelist%0D%0A exterr="ERR_CLIENT_GONE|WITH_CLIENT" 
>>>>> 
>>>>> 1715537183.180  99993 172.16.31.205 TCP_MISS_ABORTED/000 0 POST http://pjcpd-dlpend01.hlbank.my/GECommunicationWS.asmx - HIER_NONE/-:- - mac="00:00:00:00:00:00" accessrule:%20global_whitelist%0D%0A exterr="ERR_CLIENT_GONE|WITH_CLIENT" 
>>>>> 
>>>>> I have imported the access.log into my ELK machine and I can see that during the time that the users complained about the slowness there is a huge spike of NONE_ABORTED messages. 
>>>>> 
>>>>> https://i.postimg.cc/6QR79GWk/6e727e86-de3d-4f3b-bd9e-04c04052ca2e.jpg 
>>>>> 
>>>>> Now my question is: 
>>>>> 1. What can cause this kind of issue? It's a squid server issue, network (firewall, switch, router, …), or client? 
>>>>> 2. Why the number of NONE_ABORTED requests is almost 4 time more than normal request? 
>>>>> 
>>>>> Best regards 
>>>>> 
>>>>> _______________________________________________ 
>>>>> squid-users mailing list 
>>>>> squid-users at lists.squid-cache.org <mailto:squid-users at lists.squid-cache.org> 
>>>>> https://lists.squid-cache.org/listinfo/squid-users
>>>> 
>>> 
>>> _______________________________________________ 
>>> squid-users mailing list 
>>> squid-users at lists.squid-cache.org <mailto:squid-users at lists.squid-cache.org> 
>>> https://lists.squid-cache.org/listinfo/squid-users
>> 
>> _______________________________________________ 
>> squid-users mailing list 
>> squid-users at lists.squid-cache.org <mailto:squid-users at lists.squid-cache.org> 
>> https://lists.squid-cache.org/listinfo/squid-users
> _______________________________________________
> squid-users mailing list
> squid-users at lists.squid-cache.org
> https://lists.squid-cache.org/listinfo/squid-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.squid-cache.org/pipermail/squid-users/attachments/20240515/b76460e1/attachment.htm>