Sometime recently, the listserv at CUVMA became the listserv at CUVMB.
Last night, I did a review of the I-KERMIT list. Without waiting for
the results, i disconnected and went home. About 10:00 P.M. I got a call
from our operations staff saying they were getting console messages about
spool file limit exceeded for my userid. Since we don't run with spool
file limits that means that I had exactly 9999 spool files (the VM/XA
limit.) Not good!
I dialed in and they were all copies of the I-KERMIT list from various
peers. I purged all the files, but they continued to come in at the rate
of about one every two seconds. I removed some of the peer links to try
and break the loop but the major part of the loop appears to be Columbia
looping with itself. I just tried to call them on the phone, but they
don't open until 9:00.
If all sites which host the I-KERMIT list will check their listserv and
purge all rdr files requesting a review of I-KERMIT, as well as killing
all outbound copies of I-KERMIT NODEID headed for UGA I would appreciate
it.
Currently the list topology looks like this:
*
* PEERS= I-KERMIT@BNANDP11,I-KERMIT@UTORONTO,I-KERMIT@CLVM
* PEERS= I$KERMIT@CUVMA,I-KERMIT@DEARN,I-KERMIT@EB0UB011
* PEERS= I-KERMIT@RUTVM1,I-KERMIT@VTVM1,I-KERMIT@VTVM2
* PEERS= IKD@FINHUTC,I-KERMIT@HEARN,I-KERMIT@MARIST
* PEERS= I-KERMIT@TCSVM,I-KERMIT@UBVM,I-KERMIT@UGA
*
* List Topology
*
* UTORONTO CLVM HEARN
* | | |
* UBVM MARIST DEARN
* | | |
* UGA-----VTVM2-----CUVMA-----EB0UB011-----FINHUTC
* | | | |
* TCSVM VTVM1 RUTVM1 BNANDP11
*
Later today, when my reader clears out, I will get the list headers from
all sites and change CUVMA to CUVMB. I hope that will fix the problem.
All BITNET core sites should also delete all copies of I-KERMIT NODEID (or
I$KERMIT CUVMB or IKD FINHUTC) headed for node UGA.
Eric, How hard would it be to check for some number of review commands on
a given list in a given time and reject those which appear to be looping?
Harold
|
