On Thu, Dec 27, 2007 at 10:13:14AM -0500, Michael Galloway wrote:
> i'm having trouble getting a netapp nfs mount backed up to my local
> bacula server. this is bacula 2.2.6 patched. i think the problem is
> the large number of files on the netapp. the backup just slows to a
> crawl:
>
> molbio-fd Version: 2.2.6 (10 November 2007) x86_64-unknown-linux-gnu redhat
> Daemon started 25-Dec-07 08:02, 4 Jobs run since started.
> Heap: heap=1,306,624 smbytes=808,085 max_bytes=829,451 bufs=414 max_bufs=447
> Sizeof: boffset_t=8 size_t=8 debug=0 trace=0
>
> Running Jobs:
> JobId 92 Job birch.2007-12-25_16.47.08 is running.
> Backup Job started: 25-Dec-07 16:47
> Files=2,655,026 Bytes=373,175,385,406 Bytes/sec=2,508,371 Errors=0
> Files Examined=2,655,026
> Processing file:
> /birch_vol0/GC/organism/human_old/chromosome/8/contig/NT_007995.13/gene/grailexp/mrna/14.1.fna
> SDReadSeqNo=5 fd=5
> Director connected at: 27-Dec-07 10:06
>
> my mount options on the nfs mount are:
>
> birch:/vol/vol0 on /birch_vol0 type nfs
> (ro,rsize=32768,wsize=32768,tcp,addr=xxx.xxx.xxx.xxx)
>
> the job has been running nearly 48 hours and its only about a third done,
> there is around 1.1T on
> this filer and around 8 million files.
>
> anyone had any experience working with nfs backups like this? my other netapp
> nfs mount backups
> seem to run at adequate rates (20+MB/s)
>
>
getting even slower, at this rate, will take several days to get this backup
finished:
*status client=molbio-fd
Connecting to Client molbio-fd at molbio:9102
molbio-fd Version: 2.2.6 (10 November 2007) x86_64-unknown-linux-gnu redhat
Daemon started 25-Dec-07 08:02, 4 Jobs run since started.
Heap: heap=1,306,624 smbytes=794,986 max_bytes=829,451 bufs=389 max_bufs=447
Sizeof: boffset_t=8 size_t=8 debug=0 trace=0
Running Jobs:
JobId 92 Job birch.2007-12-25_16.47.08 is running.
Backup Job started: 25-Dec-07 16:47
Files=4,704,033 Bytes=425,852,376,675 Bytes/sec=1,830,591 Errors=0
Files Examined=4,704,033
Processing file:
/birch_vol0/database/iprscan/tmp/bpse_E254_04feb03/cnk_39/hmmpfam.out
SDReadSeqNo=5 fd=5
Director connected at: 28-Dec-07 09:24
====
i've checked the nfs mount, the ethernet connections on the server and client
(both at 1000Mb/s
full duplex), ran a traceroute, etc. could this be a bacula or postgres issue
(my db has grown
to over 2GB in size)? of do you reckon its just due to the netapp and number of
files? the netapp
is working, but not working very hard:
birch> sysstat 5
CPU NFS CIFS HTTP Net kB/s Disk kB/s Tape kB/s Cache
in out read write read write age
53% 601 0 0 605 20629 26112 285 0 0 5s
42% 487 0 0 439 14583 16787 0 0 0 7s
38% 437 0 0 388 12083 15080 128 0 0 9s
40% 457 0 0 415 13282 16845 94 0 0 9s
35% 371 0 0 342 11267 14839 0 0 0 9s
34% 363 0 0 321 9910 13795 205 0 0 11s
of course some of that is traffic not associated with the backup. anyway to
speed this up
or to otherwise organize the backup so that it can get down without blocking
days worth
of other backups?
-- michael
-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users
