On Wed, 2003-08-13 at 11:10, Bret Hughes wrote: > On Wed, 2003-08-13 at 08:21, John Szkudlapski wrote: > > Hi All > > > > Having a problem with INTERNAL mail here and I was wondering if anyone > > could shed some light onto the problem. > > > > Having just upgraded our 2 linux proxy servers from RHL7.3 to RHL9 it > > has just become apparent that internal mail (clients using Outlook) are > > taking upto 3 minutes to connect to our mail server (Solaris 9 - running > > sendmail). > > > > Our two proxy servers are configured in the following way > > > > Proxy 1 > > Public IP = > > Internal 1 = 192.168.1.1 > > Internal 2 = 192.168.4.1 > > > > Proxy 2 > > Public IP = > > Internal Card 1 = 192.168.2.1 > > Internal Card 2 = 192.168.3.1 > > > > Both running > > Red Hat Linux 9, > > Squid > > Dansguardian > > sendmail (for the admin accounts on the machine - no clients) > > bind-9.2.1(Caching Only Nameserver) > > > > With our mail server (mailhost.bsfc.ac.uk) having 4 internal IP numbers > > in addition to its public IP address. > > > > 192.168.1.252 > > 192.168.2.252 > > 192.168.3.252 > > 192.168.4.252 > > > > If I give a workstatino a public IP address and bypass our proxy > > server(s) I connect almost instantaneous, the same goes for any internal > > user who uses our WebBased E-Mail (horde) which is located on our mail > > server (external). Which leads me to believe the problem lies with > > squid/dns/sendmail on the proxy servers. > > > > Our mail server does not have any problems with IMAP or POP3 and works > > perfectly. I tested this by putting in a dummy server running RHL7.3 in > > place of our main mail server to eliminate our mail server - no change. > > > > Could it be an internal DNS isssue, we don't actually have our own DNS > > server here (the two proxys are configured to use caching-only), and are > > set to look at our main DNS (hosted by www.ja.net). I will be putting in > > our own DNS once I get time. > > > > yes DNS is very likley to be the culprit try adding the internal > ip/name combinations to the /etc/hosts file on the servers first and > then try a workstation or two. to see where the problem lies. Server > stuff should help the webmail immediately.
Not to suggest Bret's solution won't work for you, but I always take a more OSI-layer-centric approach. I agree that the problem points to DNS. But, rather than stabbing at various scenarios, you'd be better served to pull out tcpdump and analyze your traffic. Something like a "tcpdump -ni eth0 port 53" is going to be more useful, and will help in the future. Learning tcpdump troubleshooting should be a standard pre-req to any SysAdmin's skillset. P.S. I use eth0 as an example. I don't know which interface you'll need to debug on, it might be eth1, or something else entirely different. HTH. -- Jason Dixon, RHCE DixonGroup Consulting http://www.dixongroup.net -- redhat-list mailing list unsubscribe mailto:[EMAIL PROTECTED] https://www.redhat.com/mailman/listinfo/redhat-list
