No buffer space available
-
Hello
Our setup is pfsense 2.2.4 on a hyperv windows 2012 R2. Every 2 to 3 days there is no communication possible. Sometimes one nic is impacted, sometimes multiple nics. I nic restart mostly resolves the problem but in some cases i need to reboot the pfsense VM. I had the same problem from version 2.1 till 2.2.4.
The error i get:
[2.2.4-RELEASE][admin@pfSense]/root: ping 10.111.0.11
PING 10.111.0.11 (10.111.0.11): 56 data bytes
ping: sendto: No buffer space available
ping: sendto: No buffer space available
ping: sendto: No buffer space available
ping: sendto: No buffer space available
^C
–- 10.111.0.11 ping statistics ---
4 packets transmitted, 0 packets received, 100.0% packet loss
[2.2.4-RELEASE][admin@pfSense]/root:I read https://doc.pfsense.org/index.php/No_buffer_space_available and tried all the suggestions but i doesn't resolve the problem…
The mbuf status when a nic stops working:
[2.2.4-RELEASE][admin@pfsense]/root: netstat -m
514/3536/4050 mbufs in use (current/cache/total)
512/3034/3546/26584 mbuf clusters in use (current/cache/total/max)
512/3030 mbuf+clusters out of packet secondary zone in use (current/cache)
0/26/26/13291 4k (page size) jumbo clusters in use (current/cache/total/max)
0/0/0/3938 9k jumbo clusters in use (current/cache/total/max)
0/0/0/2215 16k jumbo clusters in use (current/cache/total/max)
1152K/7056K/8208K bytes allocated to network (current/cache/total)
0/0/0 requests for mbufs denied (mbufs/clusters/mbuf+clusters)
0/0/0 requests for mbufs delayed (mbufs/clusters/mbuf+clusters)
0/0/0 requests for jumbo clusters delayed (4k/9k/16k)
0/0/0 requests for jumbo clusters denied (4k/9k/16k)
0 requests for sfbufs denied
0 requests for sfbufs delayed
0 requests for I/O initiated by sendfileHas it something todo with the use of synthetic drivers? What else can i try to resolve the problem?
I've read disabling openvpn does sometime resolve the problem but we need openvpn…
Thanks in advance,
Lucerna -
No one any idea on how to resolve this?
-
You saw http://unix.stackexchange.com/questions/43/freebsd-how-to-fix-ping-sendto-no-buffer-space-available ,
The problem persists when you're are NOT running in a VM ?
-
yes i've read that item but it was not a solution for me.
It is indeed a VM. We also have a hardware box and we don't have that problem there.
Thanks for your feedback.
-
I think I have had the same problem recently. When i try and ping after my connection goes down i get this. A reboot of the firewall brings everything up again.
not in a VM. However i think this may be just a generic connection error as well. I too have to reboot the firewall every few hours to few days… Did you ever figure it out? i kind of feel like its a hardware problem till i saw this error and started chasing it.
-
CPU/Cores:
Each OpenVPN tunnel or session is using one CPU core, if you run many of them it could really be that
your VM has to less CPU cores assigned for that usage. A solution could be to assign more CPU cores to
that VM.RAM usage:
If I read that there is in each of your lines from the log files the word mbuf is used together with the word max.
what is standing for maximum, I would really suggest to high up the mbuf size urgently! But then, with less or
to less RAM size you will be perhaps ending up in a booting loop according to the to small amount of RAM.
A solution would be to assign more RAM to that VM and then high up the mbuf size.mbuf sizes:
I don´t really know how many queues are build and how this will be counted here in that case, but in former
days I was thinking of that this will be count likes this: each CPU core is opening or creating a queue for each
LAN port and this would be then really hard with a 10 core Intel E5-2600v3 CPU and three quad port NICs!!!
As today I know that is not the real counting, but it explains it perhaps a little bit better to other customers
and/or users. I would like to suggest to follow this tuning tips from the pfsense Doc`s page;
Tuning and Troubleshooting Network CardsGeneral about CPU & RAM usage:
pfSense is based on FreeBSD and this kernel was growing over all the years of existing and in former days
based on the technically standards and line speed it was even enough what the kernel space was delevering
or serving but since 1 GBit/s line speed is normal for a home based Internet account and more NICs then 2
or 3 are in usage as LAN ports it would be perhaps better from case to case to high up this mbuf size not
to run out of space that will be freezing then the whole system or the entire network part. Having at this
days a multi core CPU ~@3,0GHz and sufficient amount of RAM would be a good starting point to solve
around any kind of peak or workload. -
First off thank you very much for all the extremely detailed technical knowhow. I know it is more directed at the other person but I will respond anyways.
In my case, i am sitting at MBUF Usage: 2026/26584. Is it worth increasing? i think not as that is like 10%…. The machine has 2gb of ram and its using 9% of that currently.
Having said that, i have switched to 64 bit install and my install is stable for the last 2 days. fingers crossed. I dont have any vpn tunnels. Well i have an insecure pptp vpn tunnel sometimes but not really correlated to the times my connection fails. I am just a static IP connecting to another device with a static IP, (metrotik router).
