Bacula-users

[Bacula-users] Flaky performance using LTO-1 tapes

2008-10-30 11:21:00
Subject: [Bacula-users] Flaky performance using LTO-1 tapes
From: "Russell Sutherland" <russ AT quist DOT ca>
To: bacula-users AT lists.sourceforge DOT net
Date: Thu, 30 Oct 2008 10:57:16 -0400
I've been working (limping) along for several months using the
following hardware/software bacula autochanger tape setup:

Software:

# uname -a
FreeBSD bkup.utoronto.ca 7.0-RELEASE-p4 FreeBSD 7.0-RELEASE-p4 #2: Thu
Sep  4 08:58:43 EDT 2008
root AT bkup.utoronto DOT ca:/opt/obj/opt/src/sys/GENERIC  i386

# pkg_info | grep bacula
bacula-bat-2.4.3    The network backup solution (GUI)
bacula-client-2.4.3 The network backup solution (client)
bacula-docs-2.4.3   The Bacula document set
bacula-server-2.4.3 The network backup solution (server)


Hardware:

# dmesg | grep -i scsi
ahc0: <Adaptec 2940 Ultra SCSI adapter> port 0xb400-0xb4ff mem
0xfeafe000-0xfeafefff irq 16 at device 4.0 on pci3
aic7880: Ultra Wide Channel A, SCSI Id=7, 16/253 SCBs
da0: <3ware Logical Disk 00 1.00> Fixed Direct Access SCSI-0 device
sa0: <HP Ultrium 1-SCSI E32K> Removable Sequential Access SCSI-3 device
ch0: <DELL PV-122T D37r> Removable Changer SCSI-2 device

I've gotten less than stellar performance from the tape/autochanger set up:

# echo "list media Pool=Default" | bconsole
Connecting to Director backup.cns.utoronto.ca:9101
1000 OK: backup-dir Version: 2.4.3 (10 October 2008)
Enter a period to cancel a command.
list media Pool=Default
Automatically selected Catalog: MySQL
Using Catalog "MySQL"
+---------+------------+-----------+---------+-----------------+----------+--------------+---------+------+-----------+--------------+---------------------+
| MediaId | VolumeName | VolStatus | Enabled | VolBytes        |
VolFiles | VolRetention | Recycle | Slot | InChanger | MediaType    |
LastWritten         |
+---------+------------+-----------+---------+-----------------+----------+--------------+---------+------+-----------+--------------+---------------------+
|      36 | Catalog    | Error     |       1 |  95,398,772,324 |
22 |   31,536,000 |       1 |    0 |         1 | Media-backup |
2008-06-30 11:48:22 |
|     559 | Vol-1      | Full      |       1 | 160,904,991,744 |
683 |   31,536,000 |       1 |    1 |         1 | LTO-1        |
2008-08-27 05:31:50 |
|     568 | Vol-3      | Full      |       1 |  11,174,833,152 |
54 |   34,560,000 |       1 |    3 |         1 | LTO-1        |
2008-08-30 07:32:28 |
|     570 | Vol-4      | Full      |       1 |  23,107,295,232 |
106 |   34,560,000 |       1 |    4 |         1 | LTO-1        |
2008-09-05 05:55:30 |
|     572 | Vol-7      | Full      |       1 |  69,669,347,328 |
297 |   34,560,000 |       1 |    7 |         1 | LTO-1        |
2008-09-21 03:24:23 |
|     573 | Vol-6      | Full      |       1 |   3,281,596,416 |
19 |   34,560,000 |       1 |    6 |         1 | LTO-1        |
2008-09-22 14:00:12 |
|     574 | Vol-5      | Error     |       1 |  25,882,149,888 |
127 |   34,560,000 |       1 |    5 |         1 | LTO-1        |
2008-09-29 04:00:32 |
|     578 | Vol-2      | Full      |       1 |   2,737,889,280 |
19 |   34,560,000 |       1 |    0 |         0 | LTO-1        |
2008-10-28 03:53:34 |
|     579 | Vol-8      | Full      |       1 | 140,503,200,768 |
429 |   34,560,000 |       1 |    8 |         1 | LTO-1        |
2008-10-27 02:29:38 |
|     595 | Catalog-01 | Append    |       1 |     515,814,669 |
 0 |   34,560,000 |       1 |    0 |         0 | Media-backup |
2008-10-30 04:56:26 |
+---------+------------+-----------+---------+-----------------+----------+--------------+---------+------+-----------+--------------+---------------------+


Notice that the VolBytes is all over the map. Some tapes seem to
indicate "full" even after 3 Gbytes of data.

So I wanted to dig a bit deeper and performed a standard "btape fill"
exercise, using Vol-2, which "failed" after 2.7G last time
it was in service.

After several hours of writing, about 80Gbytes worth I received the
low level kernel error message:

...
ch0 at ahc0 bus 0 target 3 lun 0
ch0: <DELL PV-122T D37r> Removable Changer SCSI-2 device
ch0: 3.300MB/s transfers
ch0: 8 slots, 1 drive, 1 picker, 0 portals
Trying to mount root from ufs:/dev/ad0s1a
....
(sa0:ahc0:0:5:0): Unexpected busfree in Data-out phase
SEQADDR == 0x7c
(sa0:ahc0:0:5:0): lost device
(sa0:ahc0:0:5:0): failed to write terminating filemark(s)
(sa0:ahc0:0:5:0): tape is now frozen- use an OFFLINE, REWIND or MTEOM
command to clear this state.

And after this fact the /dev/sa0 device was no longer usable and btape
was wedged.

Any ideas? Either the Dell hardware is very flaky and can not be used
in a production service,
or I've got some scsi hardware tuning to do.

Thanks in advance.

-- 
Russell Sutherland
russ AT quist DOT ca
+1.416.696.7600

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users

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