Hi all,
My problem is the following : about each days i get I/O errors when the
backup is going on tape. Details follow. I tried to work with another robot,
with other tapes and after having cleaned the streamer. I didn't try to erase
the tapes since i'm not sure it will solve the problem and i needed the
content till now (only one cycle backup till now).
-- in the mail report :
*** A TAPE ERROR OCCURRED: [[writing file: Input/output error]].
*** PERFORMED ALL DUMPS TO HOLDING DISK.
<cut>
taper: tape DAILY11 kb 7210912 fm 94 writing file: Input/output error
-- in /var/log/amanda/Normal/amdump :
taper: writing end marker. [DAILY11 ERR kb 7210912 fm 94]
driver: result time 5825.636 from dumper0: FAILED 01-00188 ["data write:
Broken pipe"]
driver: result time 5825.636 from taper: TAPE-ERROR 00-00187 [writing file:
Input/output error]
dumper: kill index command
driver: QUITTING time 5825.682 telling children to quit
driver: send-cmd time 5825.682 to taper: QUIT
amdump: end at Tue Jun 19 02:41:59 CEST 2001
-- in /var/log/amanda/Normal/log :
SUCCESS taper obwec1 /var 20010619 1 [sec 39.029 kb 314400 kps 8055.4 {wr:
writers 9825 rdwait 0.834 wrwait 33.126 filemark 2.899}]
INFO taper tape DAILY11 kb 7210912 fm 94 writing file: Input/output error
FAIL taper oracle.freegates.net /home 0 [out of tape]
ERROR taper no-tape [[writing file: Input/output error]]
FATAL taper syncpipe_get: w: unexpected EOF
FAIL dumper oracle.freegates.net /home 0 ["data write: Broken pipe"]
sendbackup: start [oracle.freegates.net:/home level 0]
sendbackup: info BACKUP=/usr/bin/gtar
sendbackup: info RECOVER_CMD=/usr/bin/gtar -f... -
sendbackup: info end
FAIL driver oracle.freegates.net /home 0 [dump to tape failed]
At first i thought the problem was with the length of the tape, since i
usually got this error near the end of the tape. So i tried to set it down
(38500 kbytes instead of 40000 for a DLT).
I think it comes more early since i tried to compress some filesystems.
Today i read on the web :
16.1. Space wastage on an Exabyte tape. An Exabyte tape, if not operated fast
enough to make it stream, can waste lot of tape with padding. As much as 50%
can be lost. Also, make sure to use data grade tapes: fewer errors, better
performance, more mileage....
Have you a tip for this problem ?
Maybe i should do less dumps in parallel so each of them will get more
bandwidth - and a faster stream for each dump, so the flow comes fast enough.
Have you already seen this problem ? Do you have a hint or a solution ? I'm
still investigating further.
Thanks for your help, regards
Dexter
--
Knowledge-sharing and open-source content : another way to gain eternity.
Francis 'Dexter' Gois - [EMAIL PROTECTED]
System Administrator - Tiscali Belgium NV/SA
phone: +3224000839 - fax : +3224000899
