Chris Hoogendyk <hoogendyk AT bio.umass DOT edu> writes:
> Could you post your config file?
Sure, no problem. See below.
> There are a couple of things that could cause this. One example would
> be if you don't have a holding disk.
Nope, I've got a 2TB holding disk.
> If you are going direct to tape, then it won't dump in parallel
Right, actual dumps seem to happen in parallel, just not the
estimates. Which I think might be the maxdumps setting Jean-Louis
pointed out. I for some reason had overlooked that setting and it was
using the default.
> If that is your configuration, it could also contribute to your
> speed issues in other ways, for example causing "shoe-shining" on
> your LTOs, which would slow things down more. I don't think I've
> seen an answer on that question yet.
I think there's more going on with that than just amanda performance.
I think this server is completely mis-configured, I think our network
is probably suffering from the same misconfiguration, as is the NAS
we're trying to back. All three were put together by the same person
who has since left. I'm inheriting multiple messes which impact each
other significantly and it's impossible to tell the root cause of each
of the various problems.
> There could also be issues with what partitions or spindles things are
> mounted on, and contention from that perspective (referring to the
> speed issue).
Everything there is on a NAS, so technically everything is striped out
over N drives in a RAID5 array. I was thinking that you'd have
everything virtually on the same "spindle" in this case, but then it
was pointed out that we have 300 other systems NFS mounting from this
NAS. So, if one host can't read from all file systems on the NAS
simultaneously, it's more likely a problem with that host than it is
with the NAS.
I'm getting very frustrated because I just want to rip it all apart
and do it right, but we "don't have the time for that". Grrr.
--
Thanks,
Paul
Here's my config:
org "" # Subject line prefix for reports.
mailto "ops+backup AT permabit DOT com" # space separated list of recipients.
dumpuser "backup" # user to run dumps under
maxdumps 16 # The maximum number of backups from a single host\
# that Amanda will attempt to run in parallel.
inparallel 32 # maximum dumpers that will run in parallel (max 63)
# within the constraints of network bandwidth
# and holding disk space available
displayunit "g" # Possible values: "k|m|g|t"
# Default: k.
# The unit used to print many numbers.
# k=kilo, m=mega, g=giga, t=tera
netusage 1024 mbps # maximum net bandwidth for Amanda, in KB per sec
dumpcycle 7 # the number of days in the normal dump cycle
runtapes 4 # number of tapes to be used in a single run of amdump
tapecycle 10 tapes # the number of tapes in rotation
# dumpcycle * runtapes * 6
bumpsize 20 Gb # minimum savings (threshold) to bump level 1 -> 2
bumppercent 0 # minimum savings (threshold) to bump level 1 -> 2
bumpdays 1 # minimum days at each level
bumpmult 1.5 # threshold = bumpsize * bumpmult^(level-1)
etimeout 10800 # number of seconds per filesystem for estimates.
dtimeout 7200 # number of idle seconds before a dump is aborted.
ctimeout 30 # maximum number of seconds that amcheck waits
# for each client host
usetimestamps true
labelstr "^S[0-9][0-9]-T[0-9][0-9]$"
tapebufs 40 # A positive integer telling taper how many
# 32k buffers to allocate. WARNING! If this
# is set too high, taper will not be able to
# allocate the memory and will die. The
# default is 20 (640k).
tpchanger "chg-zd-mtx" # the tape-changer glue script
tapedev "/dev/nst1" # the no-rewind tape device to be used
changerfile "/etc/amanda/offsite/overland-mtx"
changerdev "/dev/sg1"
maxdumpsize -1 # Maximum number of bytes the planner will
# schedule for a run
# (default: runtapes * tape_length).
amrecover_do_fsf yes # amrecover will call amrestore with the
# -f flag for faster positioning of the tape.
amrecover_check_label yes # amrecover will call amrestore with the
# -l flag to check the label.
amrecover_changer "changer" # amrecover will use the changer if you restore
# from this device: amrecover -d changer
holdingdisk hd1 {
comment "main holding disk"
directory "/backups/amanda/offsite" # where the holding disk is
use 1700 Gb # how much space can we use on it
chunksize 1 Gb
}
reserve 0 # Don't reserve any space for degraded mode. We have
# 1.8TB of holding space, which should be able to hold a
# level 0 of just about everything.
autoflush yes # amdump will autoflush any backups on the holding disk
# The following are all DIRECTORIES, even 'infofile' which was left
# for legacy reasons...
infofile "/var/log/amanda/offsite/curinfo" # database DIRECTORY
logdir "/var/log/amanda/offsite" # log DIRECTORY
indexdir "/var/log/amanda/offsite/index" # index DIRECTORY
# Defines how the e-mail report is laied out.
columnspec
"HostName=0:9,Disk=1:20,Level=1:1,OrigKB=1:7,OutKB=1:7,DumpRate=1:7,DumpTime=1:6,TapeRate=1:7,TapeTime=1:6"
tapetype HPULT3 # what kind of tape it is (see tapetypes below)
define tapetype HPULT3 {
comment "HP LTO Ultrium 3, 400/800"
lbl-templ "/etc/amanda/DLT.ps"
length 386048 mbytes
filemark 0 kbytes
speed 56677 kps
}
define dumptype global {
comment "Global definitions"
# Include this dumptype in all other dumptypes if you want to
# inherit these settings
compress none
estimate client
holdingdisk required
index yes
program "GNUTAR"
strategy standard
tape_splitsize 10GB
}
define dumptype user {
comment "User development space - highest priority"
global
priority high
}
define dumptype server-config {
comment "Server configuration space - medium priority"
global
priority medium
}
define dumptype os-config {
comment "OS file systems - low priority - seldom changes"
global
priority low
}
|