I see some : openvpn 95093 TCP/UDP packet too large on write - Should i be worried ?

[bingo600]

One of my OpenVPN Server daemons have begun logging the below

Oct 13 11:04:09 	openvpn 	95093 	TCP/UDP packet too large on write to [AF_INET]112.121.x.x:44795 (tried=1572,max=1570)
Oct 13 11:04:09 	openvpn 	95093 	TCP/UDP packet too large on write to [AF_INET]112.121.x.x:44795 (tried=1572,max=1570)
Oct 13 11:04:05 	openvpn 	95093 	TCP/UDP packet too large on write to [AF_INET]112.121.x.x:44795 (tried=1572,max=1570)
Oct 13 11:04:05 	openvpn 	95093 	TCP/UDP packet too large on write to [AF_INET]112.121.x.x:44795 (tried=1572,max=1570)
Oct 13 11:04:01 	openvpn 	95093 	TCP/UDP packet too large on write to [AF_INET]112.121.x.x:44795 (tried=1572,max=1570)
Oct 13 11:04:01 	openvpn 	95093 	TCP/UDP packet too large on write to [AF_INET]112.121.x.x:44795 (tried=1572,max=1570)
Oct 13 11:03:57 	openvpn 	95093 	TCP/UDP packet too large on write to [AF_INET]112.121.x.x:44795 (tried=1572,max=1570)
Oct 13 11:03:57 	openvpn 	95093 	TCP/UDP packet too large on write to [AF_INET]112.121.x.x:44795 (tried=1572,max=1570) 

They come in bursts of 5..7 lines , and a bit random , 2..3 times a minute.
Is that just an info , or am i actually dropping data ?

How do i trace what would be generating this (besides tcpdump/wireshark) ?
Would a MAX MTU on the OVPN interface cure it ?

I see no erroors on the Client side (a pfSense 2.4.4-p3)
/Bingo

[kiokoman]

mtu issue, packets are dropped
try with

fragment 1450;
mssfix 1450;

custom option of openvpn

[bingo600]

@kiokoman

Thanx

Is that on/in the VPN server definitions ?
Do i set it on the Client side too , even if i don't see any issues there ?

/Bingo

[akuma1x]

@bingo600 That's the very first box under the Advanced Configuration of the pfsense OpenVPN server settings. You've got to either copy and paste it in there, or type it by hand.

Don't know if the client has to be set too, however.

[kiokoman]

i have it on both side on my site to site i think.. let me check what i have..

yes, it's mandatory to have the same settings on both side or you get this error
FRAG_IN error flags=0x2a187bf3: FRAG_TEST not implemented

[bingo600]

@kiokoman

Thank you for the excellent support

Much appeciated

/Bingo

[bingo600]

Just tried on two test firewalls.

Hint: Apply on client first , then on server.

1:
I applied the fix on the client , then the server , client disconnected then reconnected and came up.

2:
I applied the fix on server , then client disconnected and was "lost"
I saw these on the server

 Oct 14 08:43:01 	openvpn 	10921 	FRAG_IN error flags=0xfb00001d: bad fragment size
Oct 14 08:43:00 	openvpn 	10921 	FRAG_IN error flags=0xfb00001d: bad fragment size 

Came up after i did a HTTPS to Outside , and applied the client fix.

I'd recommend to have a HTTPS access to the "Outside ip" (NON VPN based) , just in case ...
Or you prob could remove the server fix , and do it in reverse order.

/Bingo