What caused "fta0: Received Overrun" message?

[권진호]

Hi all,

I am using Digital Alpha4100 566MHZ 2CPU/2048MB
Tru64 Unix 4.0E and FDDI Network, DAS Network Card.

The "fta0: Received Overrun" are recorded again in /var/adm/messages file .

The contents of "Error messages" are as following;

10:38:08 vmunix: fta0: Receive Overrun
10:38:12 last message repeated 2 times
10:40:00 vmunix: fta0: Link Unavailable
15:26:05 vmunix: fta0: Link Available
15:37:15 vmunix: fta0: Link Unavailable


Any clues? Any known workarounds?

Can anyone help me out please?

regards,
Man

[Rob Leadbeater]

Hi,

From the man page of fta:

http://tru64unix.compaq.com/docs/base_doc/DOCUMENTATION/V51B_HTML/MAN/MAN7/0103____.HTM

"The adapter's receive queue is full and packets are being dropped."

Cheers,

Rob

[Sri_10 · Accepted solution]

Two possibilities, not much horsepower in the system, or not much RAM (or available free memory is more important factor) on this system.

o If its low memory condition, check:

# vmstat

On the first line of output, if pout is other than "0", likely system is running low on memory, and it started paging/swapping. During such conditions, even kernel can't allocate large memory buckets needed for FTA controller.

o Second possibility, not enough horse-power (and not enough memory) on this system:

--- excerpt ---
> Receive data overruns indicate that data is being received by the
>FDDI adapter from the ring faster than the adapter can transfer
>the data to host memory. At some point, there is no more room on
>the adapter for incoming data so it generates this unsolicited
>error event.

Thank you for your reply.
The machine is following:
AlphaStation200 4/100
64 Mbytes memory
Firmware revision 4.2
NIC: fta0 : DEC DEFPA FDDI
DIGITAL UNIX V3.2A

This machine is Proxy Server machine. So the Load Average may be high.


> This type of thing can be caused by a heavily loaded system, so
>that it takes the adapter a longer time to acquire the bus to
>transfer data to host memory. When this problem occurs, what else
>is happening on the system? More information on the system
>configuration would also help in solving this problem.

The machine is sometime Hung Up.
I think that the machine is not cut out for Proxy Server. So I recommended that the machine should be replaced with good performance machine or should be replaced with network card ( formerly 10baseT ).

o Last, but least possibility is:

If this error occurs only after system been up and running for long periods, and doesn't occur during first few days of system being up, one possibility to look at would be, kernel memory growth/leak. This is a big subject on its own, search for this topic.

o One final comment:

We recommend in the Internet server
tuning guide to tune socket:sbcompress_threshold to 600 if
that is the case. I've enclosed a snippet from an upcoming
release of the tuning guide. You can view the current setting
of this by doing a 'sysconfig -q socket sbcompress_threshold'

-B Ryan

2.3.1.3 The mbuf Cluster Compression

The sbcompress_threshold attribute controls whether mbuf clusters
are compressed at the socket layer. By default, mbuf clusters are
not compressed (the sbcompress_threshold attribute it set to 0),
which can cause proxy servers and caching servers to consume all
the available mbuf clusters. This problem is more likely to occur
if you are using FDDI, instead of Ethernet. Refer to Section
1.5.1.2 for information about monitoring mbuf clustering.
(this means netstat -m)

To enable mbuf cluster compression, modify the sbcompress_threshold
attribute and specify a value. Packets will be copied into the
existing mbuf clusters if the packet size is less than this value.

Table 2-3: The sbcompress_threshold Attribute

Default Value: 0 bytes (mbuf compression is disabled) Recomended Value: If you have a proxy server or caching server, specify 600 for the value of the sbcompress_threshold attribute.


Thanks
Sri