[LinuxGam]

Just in case anyone else has similar problems in SBG, I am almost 100% sure it is their routers there or at least my router there! I read some interesting posts on rp_filter and problems with routers dropping the routing and that seemed just like my problem, sadly it didn't fix it.

After non stop reading and learning and trying everything, I assumed maybe I had somehow broken something in the host and re-installed the whole server (backed up my vm's obv). No firewalls, no added config, installed a VENET based brand new Debian guest and exactly the same issue... VM stays up for a period of time and then can't be accessed externally, but can be via the console or from the SSH host.. hence a problem with it coming in/out.

I tested using TCPDUMP between the host and guest and could see the traffic between them, then but when it left the host nothing back.... Then like magic a little while later with no config change it works, then later it drops etc... never any issues with my KVM or my host, so I am not even considering a hardware network card issue.

The reason I didn't straight away just move to bridged mode as Myatu suggested as I have learned more not giving up on problems than solving issues straight away and don't like to give up easily, even if there is a way out.

However, I am losing mail left right and center and can't move other stuff over, so have moved this morning to bridged and everything worked first time and I genuinely expect it to stay up (I hope to god it does!)

A lot of time wasted but a lot of things learnt. Thanks Myatu for the advice to move to bridged in the end.

Cheers

[LinuxGam]

Hi

You can check in the dedicated control panel or just do a traceroute to see if it is routed through the Vac routers, if it is then your server is in DDOS Protection.

Thanks!

[Neil]

Neil how do I check on the Control Panel if it's this DDOS thing affecting it as it seems weird to me that it is working then it's blocked, then it's working. I also can still console on and SSH from the hardware host. So it's only effecting traffic actually leaving the host.

I have also checked all internal firewall and fail2ban etc and cleared them off for testing it's not them.

Hi

You can check in the dedicated control panel or just do a traceroute to see if it is routed through the Vac routers, if it is then your server is in DDOS Protection.

[LinuxGam]

Neil how do I check on the Control Panel if it's this DDOS thing affecting it as it seems weird to me that it is working then it's blocked, then it's working. I also can still console on and SSH from the hardware host. So it's only effecting traffic actually leaving the host.

I have also checked all internal firewall and fail2ban etc and cleared them off for testing it's not them.

[LinuxGam]

Hi

Have a look at this guide, http://help.ovh.co.uk/Proxmox

That's the one i have been looking at amoungst 100's of other articles on the net for days. It's not like I can't set it up, it's just sometimes it works then sometimes it doesn't without any config changing!.

This has only happened since I switched my failovers to the new servers. I have even brought down all my firewalls to rule out other issues. On a plus note I now know way more about ProxMox as have read every article in the world on it's networking :-)

It working right now again, but I wouldn't be suprised if it just stops, without me touching any config... very strange. Lose access from the OpenVz's to the net and vice versa. However the host and KVM's still work. Very strange.

[Neil]

Hi

Have a look at this guide, http://help.ovh.co.uk/Proxmox

[LinuxGam]

Just checked their Proxmox config guide and just say this

Create a vm bridge

This configuration is not recommended for performance reasons and will therefore not covered in this guide.


Is it that big a problem?

[LinuxGam]

Hmm. Would it be possible to drop the Venet and go for Veth instead (keeping in mind you have to assign a vMAC to it, instead of the Proxmox generated one)? Seems loopy on OVH's end, particularly if there's been no config change on your end.

I guess there is no reason why I can't, easy enough to set up and seems to work fine on the Windows KVM.

Woke up this morning and everything worked after a whole night, so was happy. 10 mins later without even bothering to SSH on (just checked had a logon) Email and SSH not working....

I am convinced that it's something to do with moving the Failover IP's from one DC to another and maybe some routers have the config and some don't. So depending on how it choices to get there, depends on whether it works.

Or could it be this new DDOS thing, but can't work out how to check that in the control panel and don't have time to actually write a REST client just to check it all :-)

If I log a call will OVH charge my if they don't think it's their fault?

Thanks for your help Myatu

[Myatu]

Hmm. Would it be possible to drop the Venet and go for Veth instead (keeping in mind you have to assign a vMAC to it, instead of the Proxmox generated one)? Seems loopy on OVH's end, particularly if there's been no config change on your end.

[LinuxGam]

Traceroute to a Google IP from a guest gets to the Hardware node than no further

traceroute to 173.194.41.95 (173.194.41.95), 30 hops max, 60 byte packets
1 37.59.26.xxx (37.59.26.xxx) 0.029 ms 0.007 ms 0.006 ms
2 * * *
3 * * *
4 * * *
5 * * *
6 * * *
7 * * *
8 * * *
9 * * *
10 * * *
11 * * *
12 * *^C

[LinuxGam]

I had a professional server before running Proxmox and never had any issues with KVM or OpenVZ over venet. Ever since I moved my fail overs to point to the new server (last 3 days or more) I seem to sometimes lose access to ONLY the venet OpenVz containers.

You know when you go around in circles for so long you start forgetting whether its no even set up right. I am losing my sanity. I have even completely taken off all my IPTables firewalls now and can't get it to work.

What's really bugging me is that my email was up and running correctly on a Linux OpenVz as the server and then all of a sudden it stops without me change any config... the hardware node was still running fine and no errors as was the KVM.

I have been having intermittent problems on and off for 3 days now and have a EG server.. then I finally go to log a call and it warns me I will get charged £20 if it's not OVH's fault? Is this true?

Anyway for all you experts out there will post some info see if you can help me.

Hardware node. All the IP address/Gateway and stuff works fine as I have never touched it and have no issues direct on server or from KVM, but will post the routes

Destination Gateway Genmask Flags Metric Ref Use Iface
178.32.54.xxx 0.0.0.0 255.255.255.255 UH 0 0 0 venet0
178.32.54.xxx 0.0.0.0 255.255.255.255 UH 0 0 0 venet0
37.59.26.0 0.0.0.0 255.255.255.0 U 0 0 0 vmbr0
0.0.0.0 37.59.26.254 0.0.0.0 UG 0 0 0 vmbr0

Here are the OpenVZ guest routes, this is where I can't remember if needs anything more than this as gone around in circles.

Destination Gateway Genmask Flags Metric Ref Use Iface
0.0.0.0 0.0.0.0 0.0.0.0 U 0 0 0 venet0

Is it possible their routers are doing something loopy with the failovers as I moved datacenter and feel it's very odd they can go from working to not with no config changing on my server then start again ok a little while later with nothing changing then go down again with nothing changing

I also cannot access any DNS or internet sites (even by IP) on the OpenVz over VENET. Well, nor right now anyway :-) Unless it jumps back to working