LSTSRV-L Archives

LISTSERV Site Administrators' Forum

LSTSRV-L

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Topic: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
"Eric Thomas (CERN/L3)" <ERIC@LEPICS>
Wed, 1 Mar 89 12:38:43 GMT
text/plain (55 lines)
A  number  of  LISTSERV  maintainers   have  asked  me  to  consider  removing
LISTSERV@FRMOP11  from the  LISTSERV  backbone,  in view  of  the recent  RSCS
problems that they have experienced. After discussing the problem with several
LISTSERV maintainers, I have decided to  perform that change for the following
reasons:
 
1. FRMOP11 has installed PUT 8806 on RSCSV2 without checking the bucket first.
   As a consequence of  this, they were hit by a  highly pervasive bug causing
   valid files to be overlaid with garbage.
 
2. All  DIST2 jobs going  through FRMOP11,  which includes all  LISTS database
   update  jobs,  the VMSHARE/PCSHARE  files,  and  the distribution  of  most
   international mailing lists, are being corrupted as a result of this. LISTS
   update jobs are,  fortunately, rejected (thanks to  the checksum function),
   but they cause each affected server to  request a reship of the whole LISTS
   information from  the originating  site, which  means several  thousands of
   lines  times about  a hundred  backbone servers  just for  FINHUTC. Netwide
   SIGNOFF  and DELETE  are  rejected, causing  a syntax  error  and an  error
   message being sent back to the originator by each server.
 
3. Rather than  immediately restoring the previous version of  RSCS as soon as
   they found about the problem, and until  IBM provides them with a fix, they
   have decided  to keep the  new, buggy version  online. To prevent  too many
   files from being garbled, they have  decided to stop MAILER and LISTSERV in
   the meantime, thus  completely paralyzing the distribution  of most mailing
   lists.
 
FRMOP11  has caused  major damage  to the  network through  this irresponsible
behaviour, and  this is  more than  enough to justify  their removal  from the
LISTSERV backbone. I  hesitated to do this, because of  the significant amount
of extra traffic this would cause on the international lines. However, until I
come  to  an  agreement  with  EARN  and  the  LISTSERV  community  about  the
maintenance of  EARN LISTSERVs, *I*  am still  the person responsible  for the
smooth operation of the LISTSERV backbone, EARN  or overseas, and I am not the
kind of person  who will live with  such a major problem until  IBM provides a
fix.
 
Effective  10 minutes  ago,  LISTSERV@FRMOP11  is no  longer  on the  LISTSERV
backbone, and will not be re-added to it until (1) I come to an agreement with
EARN wherein EARN staff is responsible for deciding what EARN sites are on the
backbone, in which case it is no longer my business, or (2) Mr. Ippolito makes
a formal statement about the support  of RSCS and all related network products
at  FRMOPxx, stating  that  in  the future,  FRMOPxx  staff  shall follow  the
elementary rules of systems programming and respect of the users community. In
any case, this should provide an immediate relief to the problem and allow the
distribution of most mailing lists to resume, regardless of the fix FRMOP11 is
expecting from IBM.
 
I  would like  to suggest  that  the EARN-NOG  should discuss  the problem  of
quality of maintenance of networking tools  in international nodes, and make a
few directives to (supposedly) prevent this kind of problems from happening in
the future.
 
  Eric

ATOM RSS1 RSS2