Topic: Email/Internet Stops working

[Mortale]

Version: 7.1.3

The system runs fine for about 10 hours. Both Email and internet services stop working until I reboot the system and then everything is fine for another few hours.

When the system stops, I can still get into the Server-Manager.

[compdoc]

well, SME on its own is stable, unless youve added some software on your own thats not part of the distrib. I would concentrate on hardware - bad ram, hard drive, network card, broadband modem or router....

[Mortale]

No addons of any kind have been made.

I will run some hardware test and post back .

Thanks

[compdoc]

create a boot disc with this software, and run the extended ram tests and least once all the way through, its an excellent ram test:

http://oca.microsoft.com/en/windiag.asp

[bpivk]

create a boot disc with this software, and run the extended ram tests and least once all the way through, its an excellent ram test:

http://oca.microsoft.com/en/windiag.asp

And if it doesn't work post a bug.

[Mortale]

Memory and HD check out ok.

Now I am doing a wide range of ping tests (100,000) on both the red and green Nics. If they both come back as 100% I'm not what to do.

I will, of course, start a bug tracker, but I am not sure a bug is causing these problems.

[compdoc]

Problem is, its not a bug. Its something specific to your setup. And the bug tracker isnt meant to be used for support - in fact, they dont want us using it as one.

However, someone there (or here) might be able to give you a place to start looking.

[Mortale]

Pings on both Nics check out 100%

After things stop wotking, I find that I can get to the webmail page very fast, but login takes about 5 minutes to complete. Upon login I find no email of any kind not even junkmail.

The system was working very well for about 3 days or so. No new hardware or addons were put into place--it simply stopped functioning.

[mmccarn]

What are the hardware specs on your server, and what configuration settings have you made, and what add-ons (contribs) have been loaded?

Can you access the internet IP address from off-site after a failure (That is, is your WAN connection disconnecting for some reason)?

Can you still access the internet directly from the SME server (putty to your SME, 'ping google.com' or 'lynx google.com').  That is, has a SME service failed, or the internet connectivity from the sme itself?

Can you access off-site by IP address ('ping 64.233.167.99') Or, is it a DNS issue?

Is there anything interesting in any of the log files? (I'd look at /var/log/messages, /var/log/qpsmtpd/current, /var/log/squid/access.log)

Is your system out of disk space? (df -k)
[shame]I filled my disk last week and dns was the only service that failed -- but this prevented anything else from working...[/shame]

[Mortale]

What are the hardware specs on your server, and what configuration settings have you made, and what add-ons (contribs) have been loaded?

System is an IBM P IV 1.8ghz 512mb of ram and an 80gb HD. I have two 3com NICS one for RED and one for GREEN

No Add-ons of any kind.

Can you access the internet IP address from off-site after a failure (That is, is your WAN connection disconnecting for some reason)?

Yes I can. WAN is working.

Can you still access the internet directly from the SME server (putty to your SME, 'ping google.com' or 'lynx google.com'). That is, has a SME service failed, or the internet connectivity from the sme itself?

Checked them all and nothing seems out of place.



I can ping google from putty no packet loss.

Can you access off-site by IP address ('ping 64.233.167.99') Or, is it a DNS issue?

I can access off-site. Just really slow.

Is there anything interesting in any of the log files? (I'd look at /var/log/messages, /var/log/qpsmtpd/current, /var/log/squid/access.log)

Is your system out of disk space? (df -k)
[shame]I filled my disk last week and dns was the only service that failed -- but this prevented anything else from working...[/shame]

I have over 50gb free space.

Thanks so much for tyring to help me out. At this point I am gonna install 6.0.1 and see how long I can keep it running. I think 7.1.3 just may not work for me =( I will post back with any info I uncover just in case others have the same issue.

[smeghead]

It'd be a good idea to see how busy ya server is when this happens.

Use top or htop to check out the CPU & RAM usage on the box both after a clean boot & when it stops working as expected.

If as I suspect the system is bogged down then I would guess it may be flooded with mail which in turn invokes virus scanning & spam scanning (if turned on); both of these consume CPU resources & RAM, sometimes a fair bit as there are multiple threads running.  Run:

ps -aux | grep qmail

to see what mail processes are running (qmail-remote)

also try grepping for clamd & spamd processes to see how many of these are running.

On my server each spamd thread uses 35-45M & each clamd thread 43M

If you have qmHandle loaded you can check the mail queue to see what's in it (if anything).

HTH

[mike_mattos]

What type of internet access do you have?  I've had to reboot servers recently because both SME6 and SME7 failed to repair a lost internet connection.

In one case it followed a fire in a hydro station, another followed a storm, and the third where the ISP added a new range of IP addresses and routers.
 
This is VERY unusual, usually they self repair!

I'm suspicious that the ISP's are doing something to disconnect hackers and are not telling people what it is, just saying to reboot your firewall system.

( I know I can reset the network, but a reboot seems to do some other housekeeping and is something the on site people can handle from the admin panel )

[Mortale]

I changed the hardware and that seems to have helped my stability issue. (Over heating)

However, I can send mail with no trouble, but I do not receive any. I have no new firealls in pace.

any thought?

[mike_mattos]

If you can't receive mail, you've got a configuration issue!

I'd start at the registrar to see where your mail is being directed, check the /home/e-smith/Maildir/new directory to see if anything is arriving.

[CharlieBrady]

create a boot disc with this software, and run the extended ram tests and least once all the way through, its an excellent ram test:

http://oca.microsoft.com/en/windiag.asp

There's no need for additional software - the SME install disk has very good memory testing software included.