original message:
> I am having trouble with cron dying mysteriously and frequently. At
> first it went several weeks between deaths, but now I can't keep it
> going for even one day. I am running V2.0 (Rev 240) at the moment. I
> have an update planned to 3.0a (which I just got) soon. Is this a
> known problem? Is it fixed in 3.0a?
I received replies from the following respondents, all of whom pointed
out that this is a known bug, still active in version 3.0, relating to
running dump from cron. The most detailed was from John Speno who
included the readme's from the patch files.
Thanks to all for the help.
Kevin Houle <kevin_at_insosf1.infonet.net>
"Hellebx Knut" <bgk1142_at_bggfu2.nho.hydro.com>
John P. Speno, speno_at_cc.swarthmore.edu, Swarthmore College Computing Center
=============================================================================
Problem 1:
*********
Patch ID: OSFV30-013-3
The dump and rdump programs were aborting improperly. When they were
required to abort, they would kill a parent process outside of dump.
Therefore if dump were run from cron, dump would kill cron. This has
been corrected.
Problem 2: (CLD hpaqa6710)
*********
Patch ID: OSFV30-054-3
The (r)dump program was core dumping with SIGSEGV when an entry for an
nfs filesystem in the /etc/fstab did not begin with a backslash.
Additionally, this patch includes a correction for properly reporting
write failures.
Problem 3: (CLD hpaq93cb9)
*********
Patch ID: OSFV30-054-3
1) The (r)dump program was failing to check for write errors on remote tapes,
local and remote regular files and piped output. Instead it would ignore the
errors and continue as though they didn't occur.
2) The rdump program was always treating the remote medium as though it were
a regular file. As a result it would not create multi-volume dumpsets. When it
reached the end of the volume, rdump would simply abort. For remote Ultrix
and OSF systems it will now prompt for the next volume and thus create
multi-volume dumpsets.
Note: It has been observed that where the remote system is not a Digital Unix
system, the rdump program may report a "failure to establish connection
to the remote rmt command". In anticipation of certain circumstances
rdump has a simpler code path that can be utilized by setting a specific
environment variable:
setenv OSF_RDUMP_SIMP_RCMD
This can be done globally or on a per command basis.
=============================================================================
-----------------------------------------------------------------------
| David M. Rocke Phone: (916) 752-7368 |
| Graduate School of Management FAX: (916) 752-2924 |
| University of California, Davis E-mail: dmrocke_at_ucdavis.edu |
| Davis, CA 95616-8609 |
-----------------------------------------------------------------------
Received on Mon Jan 23 1995 - 18:43:36 NZDT