My amdump is suddenly running for days.
I have a special configuration named "Weekly" that does a level 0 dump of
all file systems (with no record) that I have been running every Friday
night. This always took until some time on Sunday, but lately it has
carried over to Wednesday or even Thursday. Yes, I have added more
file systems to dump, but how do I figure out why adding 5% to the
work to be done makes it run for 100% longer?
I have tried to read the amdump log, but I can't decode it. Are there
any tools to help me figure out where the time is being used?
Also, how do I "tune" this thing? Should I allow more parallel dumps? Or
is that self defeating?
I have been a happy user for several years, but I am now getting heat from
my users who can't get files restored because the system is too busy doing
backups.
I can supply configs and logs upon request, but a little info may be helpful:
Server: Amanda 2.4.2 on Solaris 8 with 52Gbyte holding disk
Tapes: DLT7000 on an Exabyte 230D
usually uses 5 tapes / run
Clients: 182 file system on 25 hosts with amanda 2.4.1p1 or 2.4.2
with Solaris, Unixware, Sinix, HPUX, AIX, FreeBSD
Can someone please help me make this run faster?
--
Gary Algier, WB2FWZ [EMAIL PROTECTED] +1 856 787 2758
Ulticom Inc., 1020 Briggs Rd, Mt. Laurel, NJ 08054 Fax:+1 856 866 2033
A self-addressed envelope would be addressed "envelope."
