[SOLVED] Centos 7 bug on VM?

[paralosreg]

Hello.
I have problems with network in a Centos7 installation. I have intermitent cuts in network. It work again in a minute more or less or with a "service network restart" from the Proxmox VM console. I have this problem with all the adapters, realtek, e1000 and virtio. This is my ifcfg-eth0 config:

DEVICE=eth0
BOOTPROTO=none
ONBOOT=yes
USERCTL=no
IPV6INIT=no
PEERDNS=yes
TYPE=Ethernet
NETMASK=255.255.255.255
IPADDR=XX.XX.XX.XX
GATEWAY=XX.XX.XX.XX
ARP=yes
HWADDR=XX.XX.XX.XX

Are there any known bug with CentOS 7 and VM's? This is a similar issue

 

[claudiomiguelmuller]

I had the Proxmox 3.1 on VM with Centos 4.8 that work well.
I installed this VM CentOS 4.8 on Proxmox 4.2 and now I have a similar problem

My post:
https://forum.proxmox.com/threads/problem-update-proxmox-3-1-to-4-2.28829/

Other post with this problem
https://forum.proxmox.com/threads/lxc-containers-partially-lose-inbound-packets.28982/

 

[paralosreg]

I had the same problem with Proxmox 3.X (I don't remember... 8 maybe) and this is the reason I update to Proxmox 4.2.
In 3.8, I had 2 main systems: CentOS 6 VM and CentOS 7 CT. I had problems with both. Now because I had Nginx, I had 2 VM with CentOS 6 and CentOS 7. In the first, I had seen some cuts but not often. In CentOS 7 I have cuts constantly. I need to know if It's a CentOS bug because I work with VestaCP and I need CentOS or Debian. In case of CentOS bug, I'll migrate the servers to Debian. I see nothing strange in the logs. I discard a hardware problem because I can access to Proxmox interface without problem every time (same ip range). I don't work with LXC and I check no connection with ping from the Proxmox console.

 

[LnxBil]

Try to capture the device traffic from the tap device and try to analyze it with wireshark. Maybe that yields something.

 

[paralosreg]

Ok. I have a pcap during a connection cut. What have I to look for? I don't know so much about traffic analysis

 

[paralosreg]

I just did a fresh install of Debian and have the same problems as in CentOS. Actually, I'm seriously considering migrating my server to VMware. It definitely is a problem of Proxmox

 

[LnxBil]

Hmm, I doubt that it is a Proxmox VE problem per se. I'm running literally hundreds of virtualized Debian, RHEL, OL, CentOS systems (KVM and LXC) without ever experiencing this problem.

Would you please share more information about hardware, network connection, test environment for drops, etc.

For the analysis of the pcap file, look at the time of the breakdown and see if there are more errors (e.g. checksum) than before.

 

[paralosreg]

I have a SoYouStart server. I set the connection following this OVH guide http://help.ovh.co.uk/bridgeclient . I have installed VestaCP. I have another CentOS 7 VM running with less problems ... I'm confused, the only diference between them is VestaCP. I'll try with a fresh CentOS install

 

[claudiomiguelmuller]

Hello,
in my case it seems that decided.

In the VMs that are in Proxmox 4.2 changed all network cards for E1000 and everything seems fine.

In the VMs that are in Proxmox 3.1 / 3.3 the same VMs work with Virtio network card, but the virtio does not work well in Proxmox 4.2

 

[paralosreg]

Ok. after a lot of probes, the problem is out my VM . I have no problem in another IP range of OVH. Thx