[Info-vax] PEA0: Errors

[Christoph Gartmann]

In article <goorr9$ge3$1 at south.jnrs.ja.net>, "Richard Brodie" <R.Brodie at rl.ac.uk> writes:
>
>"Christoph Gartmann" <gartmann at nonsense.immunbio.mpg.de> wrote in message 
>news:goopmg$jbo$1 at news.belwue.de...
>
>> Yes, the switch tells something about excessive collisions and drop rates.
>
>What does "show int n" on the relevant ports look like? And/or
>mc lancp show dev/count and show dev/int on the host.


  Name  :
  MAC Address      : 001db3-062dff
  Link Status      : Down
  Totals (Since boot or last clear) :
   Bytes Rx        : 1,292,132,087      Bytes Tx        : 3,447,727,258
   Unicast Rx      : 2,575,752          Unicast Tx      : 6,609,114
   Bcast/Mcast Rx  : 30,274             Bcast/Mcast Tx  : 1,033,585
  Errors (Since boot or last clear) :
   FCS Rx          : 0                  Drops Tx        : 2,654,260
   Alignment Rx    : 0                  Collisions Tx   : 117,101
   Runts Rx        : 0                  Late Colln Tx   : 0
   Giants Rx       : 0                  Excessive Colln : 0
   Total Rx Errors : 0                  Deferred Tx     : 0
  Others (Since boot or last clear) :
   Discard Rx      : 0                  Out Queue Len   : 0
   Unknown Protos  : 56,930,250
  Rates (5 minute weighted average) :
   Total Rx  (bps) : 0                  Total Tx  (bps) : 0
   Unicast Rx (Pkts/sec) : 0            Unicast Tx (Pkts/sec) : 0
   B/Mcast Rx (Pkts/sec) : 0            B/Mcast Tx (Pkts/sec) : 0
  Utilization Rx  :     0 %            Utilization Tx  :     0 %


Device Counters ESA0:
             Value  Counter
             -----  -------
             14696  Seconds since last zeroed
            467659  Data blocks received
            133864  Multicast blocks received
                 0  Receive failure
         355890484  Bytes received
          14664775  Multicast bytes received
                 0  Data overrun
            187010  Data blocks sent
              8603  Multicast packets transmitted
               424  Blocks sent, multiple collisions
              1582  Blocks sent, single collision
              1115  Blocks sent, initially deferred
          42881697  Bytes sent
           1018409  Multicast bytes transmitted
                81  Send failure
             65535  Collision detect check failure
                 0  Unrecognized frame destination
                22  System buffer unavailable
                 0  User buffer unavailable
 

Note that the machines are currently connected to 3Com switch. Initially we
thought the HP switch might be the cause. The above refers to a microVAX.
Some errors are caused by a cluster member crash in the meantime.
The Alpha ports have neither drops nor collisions, although some of them
crash more often than the VAX. On the other hand, every 10HD port shows
a huge amount of "drops", even printers or other network gadgets. So I assume
these "drops" are nothing to worry about.

>> Meanwhile I think I found the reason: the packet sniffer shows that it is an
>> antispam box that produces huge amounts of illegal tcp sequence numbers every
>> 37 minutes so that the switch gets overwhelmed
>
>Without looking at the interface configuration, I would guess you have a
>misconfiguration that shows up under high load, and the antispam box is only
>the secondary cause.

Hm, the setup was stable until we replaced the 3Com switch with one from HP.
Even then the setup was stable until we upgraded the switch firmware from 
version K13.25 to K13.45. Afterwards we discovered a similar problem. So we
went back to K13.25. Again things were stable. Two weeks ago we gave K13.51
a try and things were fine until yesterday. When the problem started we
thought about the switch and replaced it with the one from 3Com. But with no
luck...

Since we shut down the antispam box things are back to normal again. Currently
we are setting up the replacement box and see, how this one will behave...

Regards,
   Christoph Gartmann

-- 
 Max-Planck-Institut fuer      Phone   : +49-761-5108-464   Fax: -80464
 Immunbiologie
 Postfach 1169                 Internet: gartmann at immunbio dot mpg dot de
 D-79011  Freiburg, Germany
               http://www.immunbio.mpg.de/home/menue.html