Amanda-Users

Re: very slow backup over WAN

2003-04-09 20:03:40
Subject: Re: very slow backup over WAN
From: Stephen Carville <carville AT cpl DOT net>
To: Amanda Users <amanda-users AT amanda DOT org>
Date: Wed, 9 Apr 2003 15:20:03 -0700
???

5 GB (gigabyte) of data over a 512 Kb (kilobit/sec) line should take:

(5E9 Byte)(8 bit/Byte) / (512E3 bit/sec) = 78,125 sec

(78125 sec) / (3600 sec/hour) = 21.7 hour

Even with some good compression, 17 hours doesn't look bad at all.

On Wednesday April 09 2003 02:01 pm, Shawn Clark wrote:
> Hello,
>
> Im having a bit of a problem backing up over a WAN connection.
>
> The problem is that it took 17 hours to backup roughly 5G of data over a
> 512K circuit.
>
> The 2 boxes are pretty much identical in software and hardware.
>
> Debian Distro(testing)
> Amanda 2.4.3
> Freeswan 1.96/1.99
> Gnutar 1.3.25
> Kernel 2.4.20
>
> Hardware:
> AMD 1800+
> 512M RAM
> ATA100 drives(mirrored - software raid)
>
> I used `netperf` to test my bandwidth (not during the backup) and it
> averages about 610K/sec. Makes sense since both ends have a Frac T1 @
> 512K.
> During the backup netperf returns 2-6K/sec.
>
> The Amanda server side does not seem to be clogged during the backup
> process, as I can get into it remotely and there is very little latency.
> However the client side is VERY slow.
>
> When I view the process with "top" it doesn't seem like gzip & tar are
> really eating up that much processor and memory.
>
>
> Here are my configs:
>
> /etc/amanda/DailySet1/amanda.conf:
> -----
> org "DailySet1" # your organization name for reports
> mailto "mtx" # space separated list of operators at your site
> dumpuser "backup" # the user to run dumps under
>
> inparallel 4 # maximum dumpers that will run in parallel
> netusage 10000 # maximum net bandwidth for Amanda, in KB per sec
>
> dumpcycle 4 weeks # the number of days in the normal dump cycle
> tapecycle 21 tapes # the number of tapes in rotation
>
> bumpsize 20 MB # minimum savings (threshold) to bump level 1 -> 2
> bumpdays 1 # minimum days at each level
> bumpmult 4 # threshold = bumpsize * (level-1)**bumpmult
>
> runtapes 1 # explained in WHATS.NEW
>
> tpchanger "chg-multi" # the tape-changer glue script, see TAPE.CHANGERS
> changerfile "/etc/amanda/DailySet1/chg-multi.conf"
>
> tapetype HARD-DISK # what kind of tape it is (see tapetypes below)
> labelstr "^DailySet1\.[0-9][0-9]*$" # label constraint regex: all tapes
> must match
>
> diskdir "/var/backups/amanda" # where the holding disk is
> disksize 4500 MB # how much space can we use on it
>
> infofile "/var/lib/amanda/DailySet1/curinfo" # database filename
> logfile "/var/log/amanda/DailySet1/log" # log filename
>
> indexdir "/var/lib/amanda/DailySet1/index"
>
> define tapetype HARD-DISK {
> comment "Hard disk instead of tape"
> lbl-templ "/etc/amanda/DailySet1/3hole.ps"
> length 15000 mbytes
>
> } first, to insure the important disks are dumped first.
>
> define dumptype hd-dump {
> comment "Backup to hard disk instead of tape - using dump"
> holdingdisk no
> index yes
> priority high
> no-record
> }
>
> define dumptype hd-tar {
> hd-dump
> program "GNUTAR"
> comment "Backup to hard disk instead of tape - using tar"
> options compress-fast, index
> }
>
> define dumptype low-hd-tar {
> hd-dump
> program "GNUTAR"
> comment "Root partitions with compression to hard disk"
> options compress-fast, index
> priority low
> }
>
> includefile /etc/amanda/interfaces
>
> *All other dumptypes are variations like above, eg. priority med,high,etc..
> -----
>
> /etc/amanda/DailySet1/disklist:
> ------
> mtxsac.sac / {
> low-hd-tar
> exclude list "/etc/amanda/exclude_lists/root.excl"
> compress client fast
> } 0 local
>
> All DLE's have spindle number "0" at this location
>
> max.corp / {
> low-hd-tar
> exclude list "/etc/amanda/exclude_lists/root.excl"
> compress client fast
> } 1 eth0
>
> All DLE's have spindle number "1" at this location
>
> *Again other DLE's are same as above 2 examples only med-high priority
> variations.
>
> On the first run I had the spindle number set at "0" for all of the DLE's
> at both sites. I changed it according to above, but it didn't help.
> ------
>
> /etc/amanda/interfaces:
> -----
> define interface local {
> comment "a local disk"
> use 20000 kbps
> }
>
> define interface eth0 {
> comment "512 Kbps ethernet"
> use 5000 kbps
> }
> -----
>
>
> Here is my amreport output:
>
> These dumps were to tape DailySet1.01.
> The next tape Amanda expects to use is: DailySet1.03.
>
> FAILURE AND STRANGE DUMP SUMMARY:
>   matrixex.c /var lev 0 STRANGE
>
>
> STATISTICS:
>                           Total       Full      Daily
>                         --------   --------   --------
> Estimate Time (hrs:min)    0:00
> Run Time (hrs:min)        17:46
> Dump Time (hrs:min)       17:45      17:45       0:00
> Output Size (meg)        4787.7     4787.7        0.0
> Original Size (meg)      7709.6     7709.6        0.0
> Avg Compressed Size (%)    62.1       62.1        --
> Filesystems Dumped           14         14          0
> Avg Dump Rate (k/s)        76.7       76.7        --
>
> Tape Time (hrs:min)       17:45      17:45       0:00
> Tape Size (meg)          4788.4     4788.4        0.0
> Tape Used (%)              32.0       32.0        0.0
> Filesystems Taped            14         14          0
> Avg Tp Write Rate (k/s)    76.7       76.7        --
>
>
> FAILED AND STRANGE DUMP DETAILS:
>
> /-- matrixex.c /var lev 0 STRANGE
> sendbackup: start [matrixex.corp.ontrols.com:/var level 0]
> sendbackup: info BACKUP=/bin/tar
> sendbackup: info RECOVER_CMD=/bin/gzip -dc |/bin/tar -f... -
> sendbackup: info COMPRESS_SUFFIX=.gz
> sendbackup: info end
> --snip--
>
> | Total bytes written: 245749760 (234MB, 117kB/s)
>
> sendbackup: size 239990
> sendbackup: end
> \--------
>
>
> NOTES:
>  --snip--
>   taper: tape DailySet1.01 kb 4903296 fm 14 [OK]
>
>
> DUMP SUMMARY:
>                                      DUMPER STATS            TAPER STATS
> HOSTNAME     DISK        L ORIG-KB OUT-KB COMP% MMM:SS  KB/s MMM:SS  KB/s
> -------------------------- --------------------------------- ------------
> max.corp     /           0   37470  16032  42.8   4:41  56.9   4:41  57.1
> max.corp     /home       0 18085001430752  79.1 419:52  56.8 419:52  56.8
> max.corp     /opt        0      10     64 640.0   0:00   0.0   0:00 152.1
> max.corp     /safe       0   11000   3616  32.9   1:03  57.0   1:03  57.4
> max.corp     /usr        0  422140 198720  47.1  59:03  56.1  59:03  56.1
> max.corp     /var        0  239990 116160  48.4  34:09  56.7  34:09  56.7
> max.corp     /work       0 30306401834176  60.5 538:53  56.7 538:53  56.7
> mtxsac.sac   /           0   36450  17696  48.5   0:082089.3   0:082090.5
> mtxsac.sac   /home       0   61160  57984  94.8   0:144144.7   0:144142.9
> mtxsac.sac   /opt        0      10     64 640.0   0:00   0.0   0:00 224.7
> mtxsac.sac   /safe       0      10     64 640.0   0:00   0.0   0:00 473.2
> mtxsac.sac   /usr        0  415960 170400  41.0   2:001418.2   2:001418.1
> mtxsac.sac   /var        0  156850  39424  25.1   0:271440.6   0:271441.4
> mtxsac.sac   /work       0 16744301018144  60.8   4:543459.4   4:543459.1
>
> (brought to you by Amanda version 2.4.3)
>
>
>
> Any help/advice would be greatly appreciated.
>
> Thanks
>
> Shawn


<Prev in Thread] Current Thread [Next in Thread>