Hi,

We are using Proxy Squid 3 installed on a Centos 6.5 for our Internet clients. The navigation is working fine for few days and after it becomes very slow even after Proxy server retart the problem persist. When i disable caching from the router the navigation becomes normal and once enble the caching it becomes slow.
I stop caching and cleanup the cache directory and restart the caching from router, the navigation is working fine for few days and after the problem occure again.
The cache directory is not filled and server CPU and memory usage is well.

The server configuration is:
Linux hostname 2.6.32-431.3.1.el6.x86_64 #1 SMP Fri Jan 3 21:39:27 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux
Memory: 32 Gb
Cache directory:
/dev/mapper/dsk-cache1 195G 80G 106G 44 /var/cache1
/dev/mapper/dsk-cache2 146G 204M 138G 1 /var/cache2


Here below is my squid configuration:

# Leave coredumps in the first cache dir
#coredump_dir /var/spool/squid
coredump_dir /var/cache1
#AJOUT pour la TRANSPARENCE CISCO

# WCCP Router IP
wccp2_router X.X.X.X

# forwarding 1=gre 2=l2
wccp2_forwarding_method 1

# GRE return method gre|l2
wccp2_return_method 1

# Assignment method hash|mask
wccp2_assignment_method hash


# standard web cache, no auth
wccp2_service standard 0

# Fixage de l'adresse du DNS
dns_nameservers X.X.X.X 8.8.8.8

#FIN des AJOUTS

# Uncomment and adjust the following to add a disk cache directory.
#cache_dir ufs /var/spool/squid 100 16 256
cache_dir ufs /var/cache1 200000 16 256
#cache_dir ufs /var/cache2 150000 16 256

#Change
cache_mem 20000 MB
#cache_swap_low 90
#cache_swap_high 95
maximum_object_size_in_memory 1024 KB
maximum_object_size 2048 KB
half_closed_clients off
memory_pools on
memory_pools_limit 2048 MB
dns_nameservers 127.0.0.1
ipcache_size 4096
ipcache_low 90
ipcache_high 95
log_fqdn off
half_closed_clients off
cache_swap_high 100
cache_swap_low 80%

# Add any of your own refresh_pattern entries above these.
refresh_pattern ^ftp: 1440 20% 10080
refresh_pattern ^gopher: 1440 0% 1440
refresh_pattern -i (/cgi-bin/|\?) 0 0% 0
refresh_pattern . 0 20% 4320

# ADMINISTRATIVE PARAMETERS
cache_mgr XXXXXXXX
cache_effective_user squid
cache_effective_group squid
visible_hostname hostname#access_log /var/log/squid/access.log squid
acl snmpcommunity snmp_community public
snmp_port 3401
snmp_access allow snmpcommunity localhost
snmp_access deny all


Need your help as this is used by our Internet client.

Regards,

Retsa.

hmmmm... that don't look right to me... try

dns_nameservers 8.8.4.4 8.8.8.8

The problem your users are experiencing has all the hallmarks of a DNS issue.

and welcome to linuxquestions.org. As first posts go, that one was interesting.

Hi the X.X.X.X in the dns_nameservers are our DNS servers and followed by 8.8.8.8 servers.
Why are you advising to use 8.8.4.4 and 8.8.8.8.
Do you think about a problem in our DNS, but if our dns has a problem why we don't have a the same problem when we disable the Proxy caching in our router.

Those are two of googles public master DNS servers. I didn't realize X.X.X.X. was a mask, I thought it might have been in the config as a placeholder and you forgot to change it over to a live IP.
The cyclical nature of the problem. That normally means something beyond the Linux instance and the web app in question are involved (those things normally work or don't work). Resource contention on the instance or bad code in the app can cause it, but I'm not in a position to trouble shoot those for you (so I'm assuming you've already looked at those possibilities).
DNS is simple to understand as a concept however it's not nearly as simple in implementation. Zones transfer on a schedule between recursive DNS servers, the RR sigs can expire unless roller is running on both servers, which can trigger a problem on the next zone transfer...there's a lot of moving parts to it. It seems like a likely fail point based on the symptom your describing.
I don't know how much DNS caching your router is doing, nor did I know until a I read your response that you're running internal DNS servers. Squid isn't exactly a DNS resource, it caches web objects which can be anything you find on the internet, which means if your using Squid as a DNS referrer, it's searching through a ton of crap to find the name/ip keypair you're asking it for. DNS has to happen quickly or the server gets buried, which sounds similar to what your experiencing.

The DNS suggestion is just that, a suggestion, based on what I know about your problem which is a couple dozen words on an internet message board. I'm happy to bounce ideas off you but if you feel you need a professional engineering assessment, you need to hire an engineer. Linuxquestions is not a substitute for on site expertise.

Hi all,

My problme is solved, it's due to a NAT problem of my router after changing our router the Proxy is working fine.
Thank you for the help of everyboby in this forum.

regards,

Retsa.