Mail Client Connection Times

[slapshotct]

When we send mail from our mail clients (IMAP w/ SMTP), it can take 10 to 20 seconds "connecting" to Scalix to send an email. We monitored the system and it appears to be on the SMTP connection. When using telnet to connect to the server, the connection time is fast but the response time for each command can be a few seconds (ie - helo me -- wait 3 to 5 seconds and then get the 250 hello response back from the server).

I have looked through the forums and it seems that every thread with a similar issue tails off without a definitive solution. Are there any clues out there? If anyone has seen this type of issue and fixed it, please let me know how you attacked the issue!

Thanks!

-Slapshot

[dkelly]

Connection delays like that are down to an unresponsive DNS server. Please check your DNS configuration and make sure that all the nameservers that are configured actually respond in a timely manner.

Cheers

Dave

[slapshotct]

I have checked:

"host mailserver.domainname.com" on both the client machine and mail host -- both come back immediately

"dig domainname.com MX" on both the client machine and mail host -- both come back immediately

A reverse DNS entry exists for the client machine used for testing and the mail server

/etc/mail/local-host-names contains the FQDN of the server and the top level domain name (i.e. server.domain.com and domain.com)

RELAY accept 127.0.0.1 & RELAY accept .domain.com both exits in /var/opt/scalix/sys/smtpd.cfg

both name servers listed in /etc/resolv.conf reply immediately to multiple domain queries

I know I have checked a bunch of other stuff for DNS in the past that I can't think of right now, so if you can think of anything in particular, please let me know and I will give you the results.

Thanks,
-Slapshot

[slapshotct]

Since I haven't heard back, I started looking into sendmail DNS configuration documentation and found a potential change to make in the /etc/hosts and the /etc/resolv.conf files. I made the changes and still have the same issue. Here is what they currently look like:

/etc/hosts:
127.0.0.1 server.mydomain.com localhost localhost.localdomain
123.123.123.123 server.mydomain.com

/etc/resolv.conf
search localdomain
domain mydomain.com
nameserver 123.123.247.232
nameserver 134.134.158.113

Of course, the actual names and IP addresses are correct in the actual file.

Is this correct for a Scalix install? If not, what should they look like? Where else should I be looking?

[bproven]

I expereinced the same issues - what I did:

DNS: As suggested you must have internal DNS setup (with reverse). Namely your scalix server needs to respond to a forward and reverse lookup. I have mine set as mail.mycompany.local so when I nslookup mail.mycomapny.local I get the IP back as an answer AND nslookup <ip> gives me mail.mycompany.local

Test with dig or nslookup to make certain...

HOSTS/RESOLV:
Here is my hosts and resolv.conf - this works for me:

hosts:

Code: Select all

# Do not remove the following line, or various programs
# that require network functionality will fail.
127.0.0.1       localhost.localdomain   localhost
192.168.1.22    mail.mycompany.local  mail


resolv.conf:

Code: Select all

#my internal DNS server
nameserver 192.168.1.50


Hope this helps you out - I was pulling my hair out too ;)

[slapshotct]

Unfortunately, the machine responds to both the forward and reverse DNS lookups very fast. They are set up and functioning. Thanks for letting me know what you did though.. I only wished it worked for me.

Anyone else have any ideas?

[slapshotct]

I was lurking (and posting) in another thread because I thought the problems were similar. The thread is located here:

http://www.scalix.com/community/viewtopic.php?t=4755

Apparently, the user cswihart was able to resolve his problem by switching to OpenDNS instead of the DNS server on his network. I tried this and unfortunately, am still stuck.

Is there anyone that can help with this? I am starting to get the feeling this may be a known problem with Scalix 10.0.5.

[ScalixSupport]

Hi!

Why do you have the following entry in the /etc/hosts file:

/etc/hosts:
127.0.0.1 server.mydomain.com localhost localhost.localdomain

Can't this be:

127.0.0.1 localhost.localdomain localhost
<some IP> server.mydomain.com server

NB: <some IP> is the IP address that you have for server.mydomain.com

Thanks,
Subir

[slapshotct]

Sorry, that was a legacy post. I changed that line as a test, which of course didn't work, and it has since been switched back. My /etc/hosts file looks like this:

Code: Select all

127.0.0.1               localhost localhost.localdomain
123.123.123.230         mailserver.mydomain.com mailserver


Please help -- the client is about to dump Scalix in favor of another solution.

Thanks

[ScalixSupport]

How much memory is assigned to Tomcat on the server?
(check JAVA_OPTS line in /opt/scalix-tomcat/bin/setclasspath.sh)

What kind of disk configuration are you using on the server? (SW-RAID, HW-RAID, Cache, etc).

How much total memory on your server?

How many users online at the same time?

CPU configuration of server?

Thanks,
Subir

[slapshotct]

My JAVA_OPTS says this:

Code: Select all

JAVA_OPTS="-server -Djava.net.preferIPv4Stack=true"


Disk config is Hardware Raid1 for OS and Raid5 for the data partition
Total memory: 4G
There are about 10 - 15 people on at once
CPU = dual (SMP) Xeon 3.2 Ghz

This machine should be overkill.

Also, is Tomcat even a factor in omsmtpd? I just want to make sure we are going down the right path. Just to reiterate, if I telnet to this server, from a server that has a properly resolving reverse DNS, it can take 3 - 10 seconds to see the 220 message to come up, then I issue a 'helo domain.com' and wait another 3 - 10 seconds before it says helo back, and so on. At times, it will respond back faster and other times it takes longer.

Thanks,

[ScalixSupport]

Hi!

First the two things I must suggest are:

Set:
JAVA_OPTS="-server -Djava.net.preferIPv4Stack=true -Xms1024m -Xmx1024m"

Try running the command below and see if it helps:
omtidyallu -M

-M caches MIME conversion data.

Take a look at Network collision rate and if it is high try to reduce. Also, check the cable connectivity, etc.. to verify there is no hardware failure or fault.

Please reply if issue persits.

Warm Regards,
Subir

[slapshotct]

I tried the changes and it hasn't made any difference.

Any other suggestions? Any information you need from me?

Thanks.

[ScalixSupport]

Hi!

Make sure after making changes for JAVA_OPTS in the setclasspath.sh, restart scalix-tomcat service.

Also,
1) Change the /etc/nsswitch.conf file hosts line to read:
hosts files dns
2) Verify the correct name server in /etc/resolv.conf
3) Include all local hosts in the /etc/hosts file
4) Check the file /etc/host.conf, this should be
order hosts,bind

Can you please tell us the scalix version that you have?

Thanks,
Subir

[ScalixSupport]

Hi!

I did discuss this issue with my colleagues, according to them, telnet'ing to socket 25 does
not do any DNS lookups, but sending a message does. So you need to resolve the DNS
issues to help improve response time.

Thanks,
Subir