You may want to install sysstat (sar) and look at the rates by doing
sar -n EDEV and sar -n DEV to compare the drops vs packets. In my
experience on critical production systems if the rate is less than 1
(drop/error/...) per 10,000 packets then in general you won't see a
performance impact. The lower the number gets (at say 1k things can
be seen) the worse it gets. In general if you get the 1 issue/40k
packets or greater that is a pretty clean network.
On Fri, May 25, 2018 at 10:24 AM, Thomas Dineen <tdineen(a)ix.netcom.com> wrote:
> Alex:
>
> Trivial answer: Slow Server drops packets. It takes a lot of server horse
> power
>
> to process a 1GB wire speed flow of packets.
>
> Thomas Dineen
>
>
> On 5/24/2018 6:43 PM, Alex wrote:
>>
>> Hi,
>>
>> Can someone explain why an interface would start showing dropped
>> packets and overruns? I have about six machines on a local LAN (the IP
>> is associated with the br0 device), and all have at least some amount
>> of dropped packets. This is one example from one of the machines on
>> the LAN; the LAN interface on the gateway machine is very similar.
>>
>> eno1: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500
>> inet6 fe80::ec4:7aff:fe7a:73f4 prefixlen 64 scopeid 0x20<link>
>> ether 0c:c4:7a:7a:73:f4 txqueuelen 1000 (Ethernet)
>> RX packets 2294973231 bytes 1227551884960 (1.1 TiB)
>> RX errors 0 dropped 159933 overruns 2252 frame 0
>> TX packets 2707484667 bytes 1948072588485 (1.7 TiB)
>> TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0
>> device memory 0xc7200000-c727ffff
>>
>> I recently rebooted the gateway and noticed it there first. It's a
>> fedora25 system acting as a gateway with shorewall. The LAN side is a
>> 1Gbs ethernet on a gigabit switch. The WAN side is a 10mbit ethernet
>> link in a colo. I suspect this machine is the cause, as nothing's
>> changed on the LAN machines for a while, and the dropped packet count
>> isn't incrementing fast enough to coincide with greater than 1TB of
>> traffic.
>>
>> I have IPMI access to the machines on the LAN, so can do testing, but
>> I don't have IPMI access to the gateway, so can't really do much
>> without having to drive to the colo first.
>>
>> What's the typical cause of these errors? I thought it was perhaps the
>> duplex mode or other link setting, but they all appear to be the same
>> (1000/full).
>>
>> There aren't any dropped packets or overruns on the WAN interface on
>> the gateway, but could some signal or other data from the WAN side be
>> causing this?
>>
>> I can run wireshark or something similar, but it's been a while, so if
>> that's your recommendation, I'd really appreciate it if you could
>> provide specific traces you'd think were best.
>>
>> Ideas greatly appreciated.
>> Thanks,
>> Alex
>> _______________________________________________
>> users mailing list -- users(a)lists.fedoraproject.org
>> To unsubscribe send an email to users-leave(a)lists.fedoraproject.org
>> Fedora Code of Conduct:
https://getfedora.org/code-of-conduct.html
>> List Guidelines:
https://fedoraproject.org/wiki/Mailing_list_guidelines
>> List Archives:
>>
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.o...
>
> _______________________________________________
> users mailing list -- users(a)lists.fedoraproject.org
> To unsubscribe send an email to users-leave(a)lists.fedoraproject.org
> Fedora Code of Conduct:
https://getfedora.org/code-of-conduct.html
> List Guidelines:
https://fedoraproject.org/wiki/Mailing_list_guidelines
> List Archives:
>
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.o...