Backup Central

On Fri, Dec 03, 2010 at 04:29:59PM -0500, upengan78 wrote:
Oh, it looks like when I add splitdisk_buffer, it shows me below error on amcheck,

amcheck test
"/opt/csw/etc/amanda/test/amanda.conf", line 43: dumptype parameter expected
"/opt/csw/etc/amanda/test/amanda.conf", line 43: end of line is expected
amcheck: errors processing config file

spelling error split_diskbuffer != splitdisk_buffer


Relevant portion from amanda.conf

holdingdisk hd1 {
directory "/random/amandahold/test"
}

define dumptype comp-tar {
program "GNUTAR"
compress fast
index yes
record yes # Important! avoid interfering with production runs
tape_splitsize 1 Gb
#fallback_splitsize 512 MB
fallback_splitsize 2048 MB
splitdisk_buffer "/random/amandahold/test"
}

define tapetype DVD_SIZED_DISK {
filemark 1 KB
length 10240 MB
}

I should have mentioned, Amanda server has 16GB of memory, you still recommend splitdisk_buffer?

+----------------------------------------------------------------------
|This was sent by upendra.gandhi < at > gmail.com via Backup Central.
|Forward SPAM to abuse < at > backupcentral.com.
+----------------------------------------------------------------------



End of included message <<<

--
Jon H. LaBadie jon < at > jgcomp.com
JG Computing
12027 Creekbend Drive (703) 787-0884
Reston, VA 20194 (703) 787-0922 (fax)

On Fri, Dec 03, 2010 at 04:04:16PM -0500, upengan78 wrote:

Now I am thinking about, removing runtapes option completely, and add splitdisk_buffer option to the holding disk, I am confused about one thing, why you mentioned that "looks like your splits are about 0.5GB. " I have actually set tape_splitsize = 1Gb although I do have fallback_splitsize = 512MB . Do you mean I should grow this value to a higher value or you said that about tape_splitsize value?


At the time I wrote that I saw you had backed up and taped
16GB of data and they were taped in 33 parts. 16GB / 33
is about 0.5GB/part.

Your amdump run used the fallback_splitsize because you
had not specified a split_diskbuffer. If one had been
specified, data would move from holding disk to the
split_diskbuffer creating "parts" of "tape_splitsize".
If a part "failed", such as at the end of a tape, then
the part would be retried on the next tape (if available).

Without a "split_diskbuffer" the parts have to be created
in memory. In this situation the parts are sized to the
"fallback_splitsize" rather than "tape_splitsize". Typically
this is smaller than tape_splitsize to avoid using too much
memory. As you had no split_diskbuffer, the fallback_splitsize
(0.5GB) was used as the part size.

--
Jon H. LaBadie jon < at > jgcomp.com
JG Computing
12027 Creekbend Drive (703) 787-0884
Reston, VA 20194 (703) 787-0922 (fax)

Thank you Jon and everyone who have helped me understand a lot of things in Amanda. I have been able to split DLEs into multiple DLEs for the sake for distributing them in proper schedule. I also tried using inparallel and maxdump in amanda config and was successful only after configuring correct spindles in DLEs.

And yes, split_diskbuffer is correct. I am using it now which I think I ignored completely earlier.

Currently, I have scheduled 2 amanda configurations in cron for amcheck/amdump. These jobs are going fine. One is weeklyfullbackup(daily incremental(5 days) and other monthlyfull ( daily incremental ) . I am thinking about buying another storage in order to accomodate additional DLEs that I have enabled. 300G is not enough. And, yes virtual tapes option looks cost effective as well..

I will write more ..

Thanks!

Hello again,

I think I am starting to realize that something needs to be configured properly in order to Fill the tapes to their MAX capacity. I'd like some advise on that. I can create another thread for that however, it may be something to do with existing configuration so I am continuing the discussion here.

I am doing below cron jobs currently (no manual amdumps)
#Amcheck
59 19 * * 1-5 su - amanda -c "/opt/csw/sbin/amcheck -a weeklyfull"
5 20 * * * su - amanda -c "/opt/csw/sbin/amcheck -a monthlyfull"
#amdump
01 00 * * 1-5 su - amanda -c "/opt/csw/sbin/amdump weeklyfull"
15 01 * * * su - amanda -c "/opt/csw/sbin/amdump monthlyfull"

weeklyfull- amanda.conf
inparallel 4
maxdumps 4
dumpcycle 7
runspercycle 5
tapecycle 9
runtapes 3
dumpuser "amanda"
tpchanger "chg-disk" # a virtual tape changer
tapedev "file:/bk/location/amanda/vtapes/weeklyfull/slots"
changerfile "/opt/csw/etc/amanda/weeklyfull/changerfile"
labelstr "WF-.*"
#label_new_tapes "PLUTO-%%"
autolabel "WF-%%"
tapetype DVD_SIZED_DISK
logdir "/opt/csw/etc/amanda/weeklyfull"
infofile "/opt/csw/etc/amanda/weeklyfull/curinfo"
indexdir "/opt/csw/etc/amanda/weeklyfull/index"
tapelist "/opt/csw/etc/amanda/weeklyfull/tapelist"
#etimeout 600 # number of seconds per filesystem for estimates.
etimeout 3600 # number of seconds per filesystem for estimates.
#etimeout -600 # total number of seconds for estimates.
# a positive number will be multiplied by the number of filesystems on
# each host; a negative number will be taken as an absolute total time-out.
# The default is 5 minutes per filesystem.
#dtimeout 1800 # number of idle seconds before a dump is aborted.
dtimeout 3600 # number of idle seconds before a dump is aborted.
ctimeout 30 # maximum number of seconds that amcheck waits
# for each client host


holdingdisk hd1 {
directory "/random/amandahold/hold"
}

holdingdisk hd2 {
directory "/random1/amanda/holdingdisk2"
}

define dumptype comp-tar {
program "GNUTAR"
compress fast
index yes
record yes # Important! avoid interfering with production runs
# tape_splitsize 1 Gb
tape_splitsize 1024 mbytes
# fallback_splitsize 512 MB
fallback_splitsize 4096 MB
split_diskbuffer "/random/buffer"
}

define tapetype DVD_SIZED_DISK {
filemark 1 KB
length 10240 MB
}





inparallel 5
maxdumps 5
dumpcycle 30 days
runspercycle 30
tapecycle 20
runtapes 9
dumpuser "amanda"
tpchanger "chg-disk" # a virtual tape changer
tapedev "file:/bk/location/amanda/vtapes/monthlyfull/slots"
changerfile "/opt/csw/etc/amanda/monthlyfull/changerfile"
labelstr "MF-.*"
#label_new_tapes "PLUTO-%%"
autolabel "MF-%%"
tapetype DVD_SIZED_DISK
logdir "/opt/csw/etc/amanda/monthlyfull"
infofile "/opt/csw/etc/amanda/monthlyfull/curinfo"
indexdir "/opt/csw/etc/amanda/monthlyfull/index"
tapelist "/opt/csw/etc/amanda/monthlyfull/tapelist"
#etimeout 600 # number of seconds per filesystem for estimates.
etimeout 3600 # number of seconds per filesystem for estimates.
#etimeout -600 # total number of seconds for estimates.
# a positive number will be multiplied by the number of filesystems on
# each host; a negative number will be taken as an absolute total time-out.
# The default is 5 minutes per filesystem.
#dtimeout 1800 # number of idle seconds before a dump is aborted.
dtimeout 3600 # number of idle seconds before a dump is aborted.
ctimeout 30 # maximum number of seconds that amcheck waits
# for each client host


holdingdisk hd1 {
directory "/random/amandahold/hold"
}

holdingdisk hd2 {
directory "/random1/amanda/holdingdisk2"
}

define dumptype comp-tar {
program "GNUTAR"
compress fast
index yes
record yes # Important! avoid interfering with production runs
# tape_splitsize 1 Gb
tape_splitsize 1024 mbytes
#fallback_splitsize 512 MB
fallback_splitsize 4096 MB
split_diskbuffer "/random/buffer"
}

define tapetype DVD_SIZED_DISK {
filemark 1 KB
length 10240 MB
}


monthlyfull amanda.conf

inparallel 5
maxdumps 5
dumpcycle 30 days
runspercycle 30
tapecycle 20
runtapes 9
dumpuser "amanda"
tpchanger "chg-disk" # a virtual tape changer
tapedev "file:/bk/location/amanda/vtapes/monthlyfull/slots"
changerfile "/opt/csw/etc/amanda/monthlyfull/changerfile"
labelstr "MF-.*"
#label_new_tapes "PLUTO-%%"
autolabel "MF-%%"
tapetype DVD_SIZED_DISK
logdir "/opt/csw/etc/amanda/monthlyfull"
infofile "/opt/csw/etc/amanda/monthlyfull/curinfo"
indexdir "/opt/csw/etc/amanda/monthlyfull/index"
tapelist "/opt/csw/etc/amanda/monthlyfull/tapelist"
#etimeout 600 # number of seconds per filesystem for estimates.
etimeout 3600 # number of seconds per filesystem for estimates.
#etimeout -600 # total number of seconds for estimates.
# a positive number will be multiplied by the number of filesystems on
# each host; a negative number will be taken as an absolute total time-out.
# The default is 5 minutes per filesystem.
#dtimeout 1800 # number of idle seconds before a dump is aborted.
dtimeout 3600 # number of idle seconds before a dump is aborted.
ctimeout 30 # maximum number of seconds that amcheck waits
# for each client host


holdingdisk hd1 {
directory "/random/amandahold/hold"
}

holdingdisk hd2 {
directory "/random1/amanda/holdingdisk2"
}

define dumptype comp-tar {
program "GNUTAR"
compress fast
index yes
record yes # Important! avoid interfering with production runs
# tape_splitsize 1 Gb
tape_splitsize 1024 mbytes
#fallback_splitsize 512 MB
fallback_splitsize 4096 MB
split_diskbuffer "/random/buffer"
}

define tapetype DVD_SIZED_DISK {
filemark 1 KB
length 10240 MB
}


weeklyfull/slots space usage
du -hk *
1K data
10G slot1
370M slot2
5.1G slot3
5.1G slot4
1.7G slot5
1.7G slot6
33K slot7
33K slot8
33K slot9


monthlyfull/slots space usage

du -hk *
10G slots/slot1
10G slots/slot2
10G slots/slot3
10G slots/slot4
10G slots/slot5
9.2G slots/slot6
10G slots/slot7
2.9G slots/slot8
10G slots/slot9
6.0G slots/slot10
6.4G slots/slot11
10G slots/slot12
2.3G slots/slot13
10G slots/slot14
8.1G slots/slot15
8.4G slots/slot16
33K slots/slot17
33K slots/slot18
33K slots/slot19
33K slots/slot20
133G slots

I haven't provided disklist file for both backup types. However, if needed I can give you that or anything else.

I wonder what I can do to fill up these tapes to the full. May be use 'flush-threshold-dumped, flush-threshold-scheduled, or taperflush ' as they mentioned here

http://wiki.zmanda.com/index.php/How_To:Delay_writing_to_tape_for_better_tape_utilization

Upendra,

I don't recall how large your holding area is or if you end up
with multiple DLE in holding waiting for tape or not.

I've had good success with "taperalgo" and selecting the "largestfit"
option. This seems to fill my (non-spanning physical) tapes pretty
close to 100%.

If you have lots of work area and can delay your dumps until there
are mulipul DLE in the work area so the taper algorithms have more
choices there are paramters for that as well.

I believe (unless they have been removed in the last itteration)
that these relatively new parameters will help you get where you
want to go.

Note this is one of several examples in the amanda.conf file.
There are several examples to look at.

# You want to fill tapes completely even in the case of failed dumps, and
# don't care if some dumps are left on the holding disk after a run:
# flush-threshold-dumped 100 # (or more)
# flush-threshold-scheduled 100 # (or more)
# taperflush 100
# autoflush yes



On Wed, Dec 15, 2010 at 01:34:31PM -0500, upengan78 wrote:
Hello again,

I think I am starting to realize that something needs to be configured properly in order to Fill the tapes to their MAX capacity. I'd like some advise on that. I can create another thread for that however, it may be something to do with existing configuration so I am continuing the discussion here.

I am doing below cron jobs currently (no manual amdumps)
#Amcheck
59 19 * * 1-5 su - amanda -c "/opt/csw/sbin/amcheck -a weeklyfull"
5 20 * * * su - amanda -c "/opt/csw/sbin/amcheck -a monthlyfull"
#amdump
01 00 * * 1-5 su - amanda -c "/opt/csw/sbin/amdump weeklyfull"
15 01 * * * su - amanda -c "/opt/csw/sbin/amdump monthlyfull"

weeklyfull- amanda.conf
inparallel 4
maxdumps 4
dumpcycle 7
runspercycle 5
tapecycle 9
runtapes 3
dumpuser "amanda"
tpchanger "chg-disk" # a virtual tape changer
tapedev "file:/bk/location/amanda/vtapes/weeklyfull/slots"
changerfile "/opt/csw/etc/amanda/weeklyfull/changerfile"
labelstr "WF-.*"
#label_new_tapes "PLUTO-%%"
autolabel "WF-%%"
tapetype DVD_SIZED_DISK
logdir "/opt/csw/etc/amanda/weeklyfull"
infofile "/opt/csw/etc/amanda/weeklyfull/curinfo"
indexdir "/opt/csw/etc/amanda/weeklyfull/index"
tapelist "/opt/csw/etc/amanda/weeklyfull/tapelist"
#etimeout 600 # number of seconds per filesystem for estimates.
etimeout 3600 # number of seconds per filesystem for estimates.
#etimeout -600 # total number of seconds for estimates.
# a positive number will be multiplied by the number of filesystems on
# each host; a negative number will be taken as an absolute total time-out.
# The default is 5 minutes per filesystem.
#dtimeout 1800 # number of idle seconds before a dump is aborted.
dtimeout 3600 # number of idle seconds before a dump is aborted.
ctimeout 30 # maximum number of seconds that amcheck waits
# for each client host


holdingdisk hd1 {
directory "/random/amandahold/hold"
}

holdingdisk hd2 {
directory "/random1/amanda/holdingdisk2"
}

define dumptype comp-tar {
program "GNUTAR"
compress fast
index yes
record yes # Important! avoid interfering with production runs
# tape_splitsize 1 Gb
tape_splitsize 1024 mbytes
# fallback_splitsize 512 MB
fallback_splitsize 4096 MB
split_diskbuffer "/random/buffer"
}

define tapetype DVD_SIZED_DISK {
filemark 1 KB
length 10240 MB
}





inparallel 5
maxdumps 5
dumpcycle 30 days
runspercycle 30
tapecycle 20
runtapes 9
dumpuser "amanda"
tpchanger "chg-disk" # a virtual tape changer
tapedev "file:/bk/location/amanda/vtapes/monthlyfull/slots"
changerfile "/opt/csw/etc/amanda/monthlyfull/changerfile"
labelstr "MF-.*"
#label_new_tapes "PLUTO-%%"
autolabel "MF-%%"
tapetype DVD_SIZED_DISK
logdir "/opt/csw/etc/amanda/monthlyfull"
infofile "/opt/csw/etc/amanda/monthlyfull/curinfo"
indexdir "/opt/csw/etc/amanda/monthlyfull/index"
tapelist "/opt/csw/etc/amanda/monthlyfull/tapelist"
#etimeout 600 # number of seconds per filesystem for estimates.
etimeout 3600 # number of seconds per filesystem for estimates.
#etimeout -600 # total number of seconds for estimates.
# a positive number will be multiplied by the number of filesystems on
# each host; a negative number will be taken as an absolute total time-out.
# The default is 5 minutes per filesystem.
#dtimeout 1800 # number of idle seconds before a dump is aborted.
dtimeout 3600 # number of idle seconds before a dump is aborted.
ctimeout 30 # maximum number of seconds that amcheck waits
# for each client host


holdingdisk hd1 {
directory "/random/amandahold/hold"
}

holdingdisk hd2 {
directory "/random1/amanda/holdingdisk2"
}

define dumptype comp-tar {
program "GNUTAR"
compress fast
index yes
record yes # Important! avoid interfering with production runs
# tape_splitsize 1 Gb
tape_splitsize 1024 mbytes
#fallback_splitsize 512 MB
fallback_splitsize 4096 MB
split_diskbuffer "/random/buffer"
}

define tapetype DVD_SIZED_DISK {
filemark 1 KB
length 10240 MB
}


monthlyfull amanda.conf

inparallel 5
maxdumps 5
dumpcycle 30 days
runspercycle 30
tapecycle 20
runtapes 9
dumpuser "amanda"
tpchanger "chg-disk" # a virtual tape changer
tapedev "file:/bk/location/amanda/vtapes/monthlyfull/slots"
changerfile "/opt/csw/etc/amanda/monthlyfull/changerfile"
labelstr "MF-.*"
#label_new_tapes "PLUTO-%%"
autolabel "MF-%%"
tapetype DVD_SIZED_DISK
logdir "/opt/csw/etc/amanda/monthlyfull"
infofile "/opt/csw/etc/amanda/monthlyfull/curinfo"
indexdir "/opt/csw/etc/amanda/monthlyfull/index"
tapelist "/opt/csw/etc/amanda/monthlyfull/tapelist"
#etimeout 600 # number of seconds per filesystem for estimates.
etimeout 3600 # number of seconds per filesystem for estimates.
#etimeout -600 # total number of seconds for estimates.
# a positive number will be multiplied by the number of filesystems on
# each host; a negative number will be taken as an absolute total time-out.
# The default is 5 minutes per filesystem.
#dtimeout 1800 # number of idle seconds before a dump is aborted.
dtimeout 3600 # number of idle seconds before a dump is aborted.
ctimeout 30 # maximum number of seconds that amcheck waits
# for each client host


holdingdisk hd1 {
directory "/random/amandahold/hold"
}

holdingdisk hd2 {
directory "/random1/amanda/holdingdisk2"
}

define dumptype comp-tar {
program "GNUTAR"
compress fast
index yes
record yes # Important! avoid interfering with production runs
# tape_splitsize 1 Gb
tape_splitsize 1024 mbytes
#fallback_splitsize 512 MB
fallback_splitsize 4096 MB
split_diskbuffer "/random/buffer"
}

define tapetype DVD_SIZED_DISK {
filemark 1 KB
length 10240 MB
}


weeklyfull/slots space usage
du -hk *
1K data
10G slot1
370M slot2
5.1G slot3
5.1G slot4
1.7G slot5
1.7G slot6
33K slot7
33K slot8
33K slot9


monthlyfull/slots space usage

du -hk *
10G slots/slot1
10G slots/slot2
10G slots/slot3
10G slots/slot4
10G slots/slot5
9.2G slots/slot6
10G slots/slot7
2.9G slots/slot8
10G slots/slot9
6.0G slots/slot10
6.4G slots/slot11
10G slots/slot12
2.3G slots/slot13
10G slots/slot14
8.1G slots/slot15
8.4G slots/slot16
33K slots/slot17
33K slots/slot18
33K slots/slot19
33K slots/slot20
133G slots

I haven't provided disklist file for both backup types. However, if needed I can give you that or anything else.

I wonder what I can do to fill up these tapes to the full. May be use 'flush-threshold-dumped, flush-threshold-scheduled, or taperflush ' as they mentioned here

http://wiki.zmanda.com/index.php/How_To:Delay_writing_to_tape_for_better_tape_utilization

+----------------------------------------------------------------------
|This was sent by upendra.gandhi < at > gmail.com via Backup Central.
|Forward SPAM to abuse < at > backupcentral.com.
+----------------------------------------------------------------------


---
Brian R Cuttler brian.cuttler < at > wadsworth.org
Computer Systems Support (v) 518 486-1697
Wadsworth Center (f) 518 473-6384
NYS Department of Health Help Desk 518 473-0773



IMPORTANT NOTICE: This e-mail and any attachments may contain
confidential or sensitive information which is, or may be, legally
privileged or otherwise protected by law from further disclosure. It
is intended only for the addressee. If you received this in error or
from someone who was not authorized to send it to you, please do not
distribute, copy or use it or any attachments. Please notify the
sender immediately by reply e-mail and delete this from your
system. Thank you for your cooperation.

Thanks for your quick advise Brian,

I still want to post my DLEs and their sizes, may be you or someone can throw additional ideas.

weeklyfull ~ total size about 58GB

client.domain.com /export/./abc /export { 12G
comp-tar
include "./abc"
} -1


client.domain.com /export/./def /export { 4.5GB
comp-tar
include "./def"
} -1


client.domain.com /export/./ghi /export { 1.2GB
comp-tar
include "./ghi"
} 2


client.domain.com /export/./jkl /export { 1GB
comp-tar
include "./jkl"
} -1


#client.domain.com /export/./mno /export { 23GB
# comp-tar
# include "./mno"
#} 2


client.domain.com /export/./pqr /export { 14GB
comp-tar
include "./pqr"
} 2

client.domain.com /export/./stu /export { 162 MB
comp-tar
include "./stu"
} -1



Monthlyfull ~ total size about 90GB


client.domain.com /export/./vwx /export { 3.3 GB
comp-tar
include "./vwx"
} -1


client.domain.com /export/./yz /export { 1.0GB
comp-tar
include "./yz"
} -1

client.domain.com /export/OTHER/./a-m /export/OTHER { 9GB
comp-tar
include "./[a-m]*"
} -1

client.domain.com /export/OTHER/./n-z /export/OTHER { 9GB
comp-tar
include "./[n-z]*"
} 2

client.domain.com /export/OTHER/./_rest_ /export/OTHER { 9GB
comp-tar
exclude append "./[a-z]*"
} 2


client.domain.com /export/./jack /export { 2.3GB
comp-tar
include "./jack"
} 2


client.domain.com /export/./users /export { 24 GB
comp-tar
include "./users"
} -1


client.domain.com /export/./egg /export { 8GB
comp-tar
include "./egg"
} -1


client.domain.com /export/./apple /export { 21GB
comp-tar
include "./apple"
} 2


client.domain.com /export/./vls /export { 14GB
comp-tar
include "./vls"
} -1


Apart from this,

Total holding space from hd1 and hd2 is = 17 GB + 41 GB = 58GB. May be I should try to use flush-threshhold-* options if there is nothing else that can be adjusted in my case.

Ok, just an update.

I got amreport emails for weeklyfull/monthlyfull jobs that ran early this morning.

Config : weeklyfull
Date : December 16, 2010

Not using all tapes because taperflush criteria not met.
There are 1765236k of dumps left in the holding disk.
They will be flushed on the next run.

The next 3 tapes Amanda expects to use are: 3 new tapes.
The next 3 new tapes already labelled are: WF-7, WF-8, WF-9

Config : monthlyfull
Date : December 16, 2010

These dumps were to tape MF-17.
Not using all tapes because taperflush criteria not met.
There are 4129465k of dumps left in the holding disk.
They will be flushed on the next run.

The next 9 tapes Amanda expects to use are: MF-1, MF-2, MF-3, MF-4, MF-5, MF-6, 3 new tapes.
The next 3 new tapes already labelled are: MF-18, MF-19, MF-20


The config that I added to both amanda.conf (WF and MF) is below,

flush-threshold-dumped 100
flush-threshold-scheduled 100
taperflush 100
autoflush yes

From the amreport, if I understand correctly, dumped data in holding disks should get autoflushed when backup jobs run tomorrow early morning, correct? I do assume that incremental size for both jobs that are going run tomorrow will be of enough size for them to get taped..

Please let me know.

Thanks.