Intermittent Network issue from Centos 6/7 to Debian

MrCervelo · 09-06-2016, 11:13 PM

Hi All

I hope that this is the right area for this question. I'm experiencing a weird network issue with Centos 6 and 7. I am trying to connect to a Debian machine running Jira and Nagios via http and ssh and it times out. But, this is not a physical network issue as other machines (Windows) can connect to the same device with no issues. Additionally, if I run a VM (Windows 10) on the same machine, it works with no issues, and VMs running Centos 6 on Windows 10 host are experiencing the same timeout/connectivity issues.

To add even more fun to the mix, the problem is intermittent. Myself and another employee have been comparing notes, and sometimes it will work for him and not for me, and vice versa but more often than not, it will not work for either of us.

Ping and Telnet (even to port 22 and 80) work without an issue, but telnet, curl etc are all intermittent (reliably so, they all work, or they all fail).

Through another server, I was able to ssh to the destination server and I performed a tcpdump locally and on the server while I tried to ssh to it. The results were unexpected. Both showed traffic, so the data is getting through, but only the very first packet, after that both dumps show "TCP Retransmission".

I'm running out of ideas here, so any assistance is appreciated.

MensaWater · 09-08-2016, 07:40 AM

Is it possible you have more than one machine inadvertently using the same IP?

How many NICs on the source CentOS system are configured? What is the default route set to when you look at "netstat -rn"? What is in your /etc/sysconfig/network-scripts/ifcfg-<nic> files?

CentOS6/7 use Network Manager so there may be things setup in the ifcfg files that are different as compared to what CentOS5 used.

MrCervelo · 09-08-2016, 04:37 PM

Quote:

Originally Posted by MensaWater

Is it possible you have more than one machine inadvertently using the same IP?

Unlikely, the issue is affecting multiple machines and it only seems to affect this particular device. I can bypass the issue by using an SSH tunnel to another device that then connects on port 80. I shouldn't be able to tunnel if I have an IP address conflict.

Quote:

Originally Posted by MensaWater

How many NICs on the source CentOS system are configured?

I have 2 but other users only have a single NIC. However, only one NIC is actually connected to the network.

Quote:

Originally Posted by MensaWater

What is the default route set to when you look at "netstat -rn"?

Code:

netstat -rn
Kernel IP routing table
Destination     Gateway         Genmask         Flags   MSS Window  irtt Iface
0.0.0.0         192.168.111.22  0.0.0.0         UG        0 0          0 enp4s5
192.168.111.0   0.0.0.0         255.255.255.0   U         0 0          0 enp4s5
192.168.122.0   0.0.0.0         255.255.255.0   U         0 0          0 virbr0

Quote:

Originally Posted by MensaWater

What is in your /etc/sysconfig/network-scripts/ifcfg-<nic> files?

Code:

cat /etc/sysconfig/network-scripts/ifcfg-enp4s5 
TYPE="Ethernet"
BOOTPROTO="dhcp"
DEFROUTE="yes"
IPV4_FAILURE_FATAL="no"
IPV6INIT=no
IPV6_AUTOCONF="yes"
IPV6_DEFROUTE="yes"
IPV6_PEERDNS="yes"
IPV6_PEERROUTES="yes"
IPV6_FAILURE_FATAL="no"
NAME="enp4s5"
UUID="23a26e53-1f0a-4548-a2b8-a29277f1cef2"
DEVICE="enp4s5"
ONBOOT="yes"
DNS1=192.168.111.22
DNS2=8.8.8.8
PEERDNS=no
PEERROUTES=yes

Thanks for thinking about this. It was less problematic yesterday but still affecting myself and another heavy user of this system.