Hello all !
I use this distribution since 4.12. Both at home and work. I had an issue some 12 months ago with the server-manager not being accessible except from localhost either logging in at the sme machine or by remote login via ssh. Only now I found out what was the issue...
I installed the server and updated it behind the old installation of sme server, and used another internal IP, than the one in use by the gateway, of course, say 10.0.0.200 instead of 10.0.0.1 to access the server-manager. Everything went fine, then I restored a back-up from the active server and replaced it with the freshly installed machine. After the issues with the nic's and their drivers, normal after a restore on another machine, then, surprise, I could not access server-manager although the masquerading worked fine, squid, i could ping outside hosts, etc.
Restoring the old internal ip 10.0.0.200 brought server-manager back, but I had to reassign all hosts as I use static IPs and not DHCP.
I am not an expert in the templated stuff in this distribution, but I assumed it has something to do with the internal IP variable, as some services use the new one and some not.
netstat -nat does not reveal anything out of the ordinary, everything seems to listen for the right range of IPs and on the right port.
POP3 did not work either on the internal network nor SMTP. Probably there are others. I tried all I could think of, all that was possible without messing with the command line, meaning:
I swapped the NICs from the admin console (swapped<->normal mode), and then back, I put the install-time IP back and then reverted to the original from the back-up file, I changed drivers for the nics and then back again, in the hope that some of these actions could update that variable in all packages installed, but it didn't work.
Finally, I installed using the 10.0.0.1 internal IP and it works.
I hope this info will give you some hints about what is going on when doing that particular install and restore. That could possibly fix other bizarre issues, maybe.
Now, on a different note.
For quite some time I seen strange drops in internet performance, I figured it was someone sending large mails, but then unprioritized smtp traffic and the problem did not dissappear.
Netstat reveals tens, sometimes close to 100 UDP established connections on the 53 port to many DNS IPs. TLDs and such. I checked some of them and indeed resolved to something like dnsX.something.com, so there was not some kind of attack.
The problem got so bad that I had almost 100 % UDP traffic on port 53 and ICMP traffic the dest unreachable type. Small packets (couple hundred bytes each the largest ones), but added up to 20 MB/min, which is about my bandwidth limit. TCP dropped to virtually nothing and pings to my provider's gateway which take about 3-4 ms, now were over 15 seconds.
Temporarely, I am using my provider's dns, which sorta fixed the problem. I mean, although I am not behind a firewall, I specified my provider's dns in the admin console. Now things are back to normal.
Any ideas ?
I am available for any question you may have.
Thanks for the help and for such a great tool !
1 Replies
1053 Views