SUMMARY: must reboot backup server due to hung rmt process

Andreas Luik (luik@isa.de)
Tue, 07 Apr 1998 16:20:32 +0200 (MET DST)

[I asked what to do, if, after a nighly backup, a rmt process persists
on the backup server. This process is not killable, even with kill -9,
so the backup server must be rebooted to release the tape again.
System environment is: SPARCserver 20, Solaris 2.4, HP SureStore
12000e DAT autoloader.]

I received answers from:

Chris Marble <cmarble@orion.ac.hmc.edu> recommended to update to
Solaris 2.5.1 or later, because they had this problem if media
errors where found on the tape under Solaris 2.3 and 2.4, but
since 2.5.1 it works. Also suggested to try to modunload the tape
driver to regain control of the tape drive. -- [Did not yet tried
modunload, but we will consider upgrading to Solaris 2.6.]

bismark@alta.Jpl.Nasa.Gov (Bismark Espinoza) suggested to check scsi
kernel entries in www.hp.com. -- [I don't think that HP has its
own Solaris driver for this device; especially because the
documentation in <URL:http://www.hp.com:80/tape/c_sun.html> does
not mention a special driver, but explains how to configure the
Sun driver. But we will recheck our driver configuration (switch
settings) and driver configuration as suggested in this web page.]

clg@cdhg.psu.edu (Craig L. Gruneberg) sent a "mee too".

gibian@stars1.hanscom.af.mil (Marc S. Gibian) thinks that there is no
way to prevent the problem, except to use a high level backup
solution like Legato or EMC's. -- [We will consider this,
especially if the problem persists with Solaris 2.6.]

Thanks for your answers,

-- 
Andreas Luik         E-Mail: luik@isa.de
(postmaster@isa.de)  WWW:    http://www.isa.de/~luik
                     PGP:    E2 6A 41 70 67 1E 0B 68  94 0D 9E 83 95 16 AF 59