We recently deployed squid-2.6.STABLE6-5.el5_1.2 on CentOS. It ran fine for 23 days or so and began to experience time outs while resolving some websites. In the access_log entries appeared with DIRECT/www.example.com in lieu of the usual DIRECT/123.123.123.123. It only affected a couple internal servers in the morning but by afternoon google would not come up, msn continued to work. nslookups on the box to the nameservers in squid.conf worked fine. In the Cache manager -> internal dns stats, the queue grew quite big. I installed the caching nameserver package and modified the config to use bind on the local machine, this did not fix it. I ended up restarting squid (loosing 690,000 or so objects in memory). I really do not want to cron a restart, I assume that squid should keep running without these kind of problems.
On a side note we have been using Cacti to monitor squid using the built-in snmp functionality and see no results when the number of clients exceeds 500. Could this be related? Does anyone have any suggestions on how to fix this or where to begin looking to find the cause? Thanks Larry Wickham Systems Operations Spec. Eastern New Mexico University Confidentiality Notice: This e-mail, including all attachments, is for the sole use of the intended recipient(s) and may contain confidential and privileged information as defined under FERPA. Any unauthorized review, use, disclosure or distribution is prohibited unless specifically provided under the New Mexico Inspection of Public Records Act. If you are not the intended recipient, please contact the sender and destroy all copies of this message
