Amanda-Users

Re: Debugging lufs vtape and amanda flushing-throtle for amanda

2005-05-23 06:30:01
Subject: Re: Debugging lufs vtape and amanda flushing-throtle for amanda
From: Vlad Popa <vlad.popa AT sbg.ac DOT at>
To: amanda-users AT amanda DOT org
Date: Mon, 23 May 2005 12:05:21 +0200
Dear Paul,


Paul Bijnens schrieb:

Vlad Popa wrote:

So how can I say "Not so fast, Mrs. Amanda !" when taping ?


Are you 100% sure that is the problem??
The last time, it was just a simple DLE too large to fit on the
vtape.

Tape is plenty enough.

I think it must be the ftp timeout otherwise I'm clueless . See the report of today. It is strange :


These dumps were to tape daily4.
*** A TAPE ERROR OCCURRED: [[writing file: short write]].
Some dumps may have been left in the holding disk.
Run amflush to flush them to tape.
The next tape Amanda expects to use is: daily5.

FAILURE AND STRANGE DUMP SUMMARY:
 h50234     /opt/bdc lev 0 FAILED [out of tape]


STATISTICS:
                         Total       Full      Incr.
                       --------   --------   --------
Estimate Time (hrs:min)    0:00
Run Time (hrs:min)         0:28
Dump Time (hrs:min)        0:00       0:00       0:00
Output Size (meg)          25.5       25.4        0.2
Original Size (meg)        25.5       25.4        0.2
Avg Compressed Size (%)     --         --         --    (level:#disks ...)
Filesystems Dumped            9          7          2   (1:1 2:1)
Avg Dump Rate (k/s)      4937.6     5089.2      846.6

Tape Time (hrs:min)        0:13       0:11       0:01
Tape Size (meg)             2.7        2.6        0.2
Tape Used (%)               0.1        0.1        0.0   (level:#disks ...)
Filesystems Taped             6          4          2   (1:1 2:1)
Avg Tp Write Rate (k/s)     3.6        3.8        2.0

USAGE BY TAPE:
 Label        Time      Size      %    Nb
 daily4       0:13     2790k    0.1     6


NOTES:
 planner: Full dump of h50234:/home/diana promoted from 6 days ahead.
 planner: Full dump of h50234:/home/albin promoted from 6 days ahead.
 planner: Full dump of h50234:/etc promoted from 6 days ahead.
 planner: Full dump of h50234:/home/sawa promoted from 6 days ahead.
 planner: Full dump of h50234:/home/oeko promoted from 6 days ahead.
 planner: Full dump of h50234:/home/aurel promoted from 6 days ahead.
 planner: Full dump of h50234:/opt/bdc promoted from 6 days ahead.
 taper: tape daily4 kb 4128 fm 7 writing file: short write
 driver: going into degraded mode because of tape error.


DUMP SUMMARY:
                                    DUMPER STATS            TAPER STATS
HOSTNAME     DISK        L ORIG-kB OUT-kB COMP% MMM:SS  KB/s MMM:SS  KB/s
-------------------------- --------------------------------- ------------
h50234       /etc        0    2440   2440   --    0:03 930.1   9:26   4.3
h50234       /home/albin 0   12290  12290   --    0:0110680.8   N/A   N/A
h50234       /home/aurel 0      10     10   --    0:00 144.3   0:28   0.4
h50234       /home/diana 0      10     10   --    0:00 141.3   0:28   0.4
h50234       /home/oeko  0     170    170   --    0:001977.3   1:02   2.7
h50234       /home/sawa  0    8390   8390   --    0:018930.7   N/A   N/A
h50234       /home/vlad  2     100    100   --    0:001120.5   0:48   2.1
h50234       /opt/bdc    0    2650   2650   --    0:0016052.9  FAILED  ----
h50234       /opt/htdocs 1      60     60   --    0:00 599.0   0:33   1.8

(brought to you by Amanda version 2.4.5)

(BTW: I'll do the formating the output next time, but 1st she'll have to flush it all without nagging around ..)

Today the directory /opt/bdc is omitted. Tomorrow it will be another. The data amount in the DLE should fit triple on the tape.
I never saw any indication it was the vtape timing out or being too fast.

As sidenote, two weeks ago, one of my tapes was stuck during the
weekend run, and it hang amanda taper from saterday to monday morning.
When I got in mondaymorning and pressed the buttons from the tapedrive
amanda proceeded with the tapewriting as if nothing happened.
The problem happened in the beginning, when rewinding, so I can't
say it behaves the same when a write to tape would take a long time.
I never saw any indication it was the vtape timing out or being too fast.


I think in my situation it hapens during writing (quite in the middle of it ..) In YOUR case the start of writing was delayed about several days. But during writing your data stream was surely constant.

O.K...

1)So IS there a parameter to slow down the data stream to a constant level (in amanda.conf) while flushing or not ?

2) I have the 8 shares(ftp-directories) mounted on several mount points (because of the virtual tape slots) during all the time. If it is not necessary to have all "slots" mounted all over the time (is it ?) I'd try to have only the appropriate vtape (one share a time ) mounted. After use I' d "eject (aka unmount the share) . I hope, I could get more performance on the lufs filesystem by this way .

Since I have only the ftp protocol as access to the backup server I have to use lufs or ftpfs( which is outdated and followed by lufs)

Any (other) ideas to get my dumps on vtape  and maximise performance ?

Greetings

Vlad