Protocol error: uncounted data discarded

[Bob Bowen]

We just updated to 1.11.2 (from 1.10.8; both client and server are Red Hat) and
are now occasionally seeing the error:

Protocol error: uncounted data discarded

These factors seem to be involved:
1) someone checks out a large number of files (thousands)
2) when the system is busy (load average >3)
3) using a newly set tag

and the result is that files are missing from the resulting workarea. I'm
working to get more resources for our server machine, but would feel better if
there's a more definitive solution.

In searching the archives, I found a few reports of this reported in the past
few years (links and synopses (sp?) below), but nothing conclusive or
particularly helpful. Has anyone made any progress on this?

Thanks for any help,

=Bob=


http://www.mail-archive.com/address@hidden/msg16744.html

  Larry Jones says it's hard to repeat and appears to be harmless.
  
  (it's harmless as long as you don't need all your files
   the good news is it seems to work OK the second time)

http://mail.gnu.org/pipermail/bug-cvs/2001-November/006607.html

  out of memory condition?

http://www.geocrawler.com/archives/3/383/2001/8/0/6408777/

  crucial code removed?

http://www.mail-archive.com/address@hidden/msg12418.html

  fails in script, but works by hand.

http://www.mail-archive.com/address@hidden/msg12421.html

  fails when network busy

http://www.mail-archive.com/address@hidden/msg12468.html

  retry never fails

http://www.mail-archive.com/address@hidden/msg10094.html

  1.11 worked, but 1.11.1p1 fails  (related FIXME comment in server.c?)



Bob Bowen  address@hidden  Process Engineering  (952)876-4635