Re: [squid-users] Squid stops handling requests after 30-35 requests

Hi,

Please find below packet dump taken at one of the intermediate
machines on network while processing the request.

108 1.186439 10.0.11.22 165.254.58.18 TCP 76 38682 > http [SYN] Seq=0
Win=14360 Len=0 MSS=1436 SACK_PERM=1 TSval=123422360 TSecr=0 WS=64
109 1.254231 165.254.58.18 10.0.11.22 TCP 76 http > 38682 [SYN, ACK]
Seq=0 Ack=1 Win=14480 Len=0 MSS=1460 SACK_PERM=1 TSval=1476780015
TSecr=123422360 WS=2
110 1.254765 10.0.11.22 165.254.58.18 TCP 68 38682 > http [ACK] Seq=1
Ack=1 Win=14400 Len=0 TSval=123422377 TSecr=1476780015
111 1.273058 10.0.11.22 165.254.58.18 HTTP 302 GET / HTTP/1.1
112 1.273696 165.254.58.18 10.0.11.22 HTTP 1121 HTTP/1.1 503 Service
Unavailable (text/html)
113 1.273706 165.254.58.18 10.0.11.22 TCP 56 http > 38682 [FIN, ACK]
Seq=1066 Ack=235 Win=450 Len=0
114 1.274300 10.0.11.22 165.254.58.18 TCP 68 38682 > http [ACK]
Seq=235 Ack=1066 Win=17216 Len=0 TSval=123422382 TSecr=1476780015
115 1.275840 10.0.11.22 165.254.58.18 TCP 68 38682 > http [FIN, ACK]
Seq=235 Ack=1067 Win=17216 Len=0 TSval=123422382 TSecr=1476780015
116 1.275949 165.254.58.18 10.0.11.22 TCP 56 http > 38682 [RST]
Seq=1067 Win=538 Len=0
938 10.795716 10.0.11.22 165.254.58.18 TCP 76 [TCP Port numbers
reused] 38682 > http [SYN] Seq=0 Win=14360 Len=0 MSS=1436 SACK_PERM=1
TSval=123424762 TSecr=0 WS=64

From the dump when "TCP Port numbers reused" message is shown, Squid
is hanging at the same time.

Please help us on this.

Thanks,
Bhagwat

On Mon, Nov 25, 2013 at 12:36 PM, Bhagwat Yadav
<yadav.bhagwat87_at_gmail.com> wrote:
> Hi,
>
> Upgraded squid to 3.1.20-2.2 from debian.org. Issue still persists.
> Note: I have not disable the stats collection as mentioned in earlier mails.
>
> Please suggest how to resolve this?
>
> Thanks,
> Bhagwat
>
> On Fri, Nov 22, 2013 at 4:47 PM, Eliezer Croitoru <eliezer_at_ngtech.co.il> wrote:
>> And what does this 503 page content??
>> I do not know what the issue in hands is but there are couple things to
>> first test before running into full debug or try to fix issues that might
>> not exists.
>>
>> The version upgrade is there for a reason.
>> I do know why an upgrade might not solve the issues but still if you have
>> testing environment try to make sure what are the results with the latest
>> 3.1.X branch which should be 3.1.21 if I am not wrong.
>>
>> It is very critical for you to test it.
>> Since squid can run on many OS and many specs your logs are nice but not
>> helping to understand the whole issue.
>>
>> There are many bugs that was fixed from the 3.1 list but I have used it for
>> a very long time.
>>
>> If you need help installing 3.1.21 or any newer version I can try to assist
>> you.
>> Also it can be installed alongside another version.
>>
>> Best Regards,
>> Eliezer
>>
>>
>> On 21/11/13 09:38, Bhagwat Yadav wrote:
>>>
>>> Hi Eliezer/All,
>>>
>>> Thanks for your help.
>>>
>>> PFA log snippets.
>>> Log1.txt is having sample 1 of cache.log in which you can find the time
>>> gap.
>>> Log2.txt is having sample 2 of client output and cache.log showing the
>>> time gap.
>>>
>>> It seems that there is some in memory operation "StatHistCopy" which
>>> is causing this issue, not sure though.
>>>
>>> Squid version is: Squid Cache: Version 3.1.6.
>>>
>>> Please let me know that if these logs are helpfull.
>>>
>>>
>>> Thanks & Regards,
>>>
>>> On Wed, Nov 20, 2013 at 6:11 PM, Eliezer Croitoru <eliezer_at_ngtech.co.il>
>>> wrote:
>>>>
>>>> Hey,
>>>>
>>>> Can you try another test?
>>>> It is very nice to use wget but there are couple options that needs to be
>>>> consider.
>>>> Just to help you if was not there until now add: --delete-after
>>>> to the wget command line.
>>>>
>>>> It's not related to squid but it helps a lot.
>>>> Now If you are up to it I will be happy to see the machine specs and OS.
>>>> Also what is "squid -v" output?
>>>>
>>>> Can you ping the machine at the time it got stuck? what about tcp-ping or
>>>> "nc -v squid_ip port" ?
>>>> we need to verify also in the access logs that it's not naukri.com that
>>>> thinks your client is trying to covert it into a DDOS target.
>>>> What about trying to access other resources?
>>>> What is written in this 503 response page?
>>>>
>>>> Eliezer
>>>>
>>>>
>>>> On 20/11/13 12:35, Bhagwat Yadav wrote:
>>>>>
>>>>>
>>>>> Hi,
>>>>>
>>>>> I enable the logging but didn't find any conclusive or decisive logs
>>>>> so that I can forward you.
>>>>>
>>>>> In my testing, I am accessing same URL 500 times in a loop from the
>>>>> client using wget.
>>>>> Squid got hanged sometimes after 120 requests ,sometimes after 150
>>>>> requests as:
>>>>>
>>>>> rm: cannot remove `index.html': No such file or directory
>>>>> --2013-11-20 03:52:37--http://www.naukri.com/
>>>>> Resolvingwww.naukri.com... 23.72.136.235, 23.72.136.216
>>>>> Connecting towww.naukri.com|23.72.136.235|:80... connected.
>>>>>
>>>>> HTTP request sent, awaiting response... 503 Service Unavailable
>>>>> 2013-11-20 03:53:39 ERROR 503: Service Unavailable.
>>>>>
>>>>>
>>>>> Whenever it got hanged, it resumes after 1 minute e.g in above example
>>>>> after 03:52:37 the response came at 03:53:39.
>>>>>
>>>>> Please provide more help.
>>>>>
>>>>> Many Thanks,
>>>>> Bhagwat
>>>>
>>>>
>>>>
>>
Received on Mon Nov 25 2013 - 09:24:25 MST