Re: Amrecover hangs after restore
Gene Heskett <[email protected]>
| Newsgroups | gmane.comp.archivers.amanda.user |
|---|---|
| Organization | none,nada,zip |
| Message-ID | <[email protected]> |
On Wednesday 16 December 2020 16:05:15 Nathan Stratton Treadway wrote:
> On Wed, Dec 16, 2020 at 15:37:32 -0500, Gene Heskett wrote:
> > On Wednesday 16 December 2020 12:23:28 Nathan Stratton Treadway wrote:
> > > On Wed, Dec 16, 2020 at 09:42:47 -0500, Gene Heskett wrote:
> > > > You reminded me of that, so its now done. We'll see if that
> > > > fixes it.
> > >
> > > (Note that putting in the quote characters should prevent the
> > > shell from aborting due to the syntax error, but it won't fix the
> > > underlying problem that the contents of the PARTS_WRITTEN variable
> > > appear to be bogus at that point in time. Though if you want to
> > > debug that issue further, it's probably best if you reply to that
> > > branch of this thread directly :) )
> >
> > I'll see if, in my somewhat decreased mental state, I can figure out
> > how to echo that into the log file. Jon L. seems to think a bit of
> > perl that
>
> Well, there was a tantalizing hint already included in the output you
> posted to the list, so you may not need to actually chage the script
> for that part.
>
> Perhaps you shoud just go ahead and post the script; that would at
> least let us see what what script processing matches up with the
> output you posted earlier.
>
That was probably 70% of that script. But perms problems won't let me
directly paste insert it here. But mc will copy it to ~/gene/bin and
change the ug bits.
So here is the that script------
#!/bin/bash
# this script is called by backup.sh (flush.sh too) to append the
# current indices and configs in effect at the time amdump was last run
# affording a good chance of being able to do a bare metal recovery
# somewhat less painfully than they normally are.
# script Copyright 2006 Gene Heskett, license=GPLv2
# Read config
. /GenesAmandaHelper-0.61/gene.conf
PATH=/bin:/usr/local/sbin:/usr/bin
#INDICE_PATH=$INDICE_PATH/$1
echo INDICE_PATH in bak-indice-configs now=$INDICE_PATH
# 2010/09 recently amanda has stopped using the 'data' link as a pointer to the right 'tape', so...
#if [ $# -ne 1 ] ; then
if [ $# -ne 2 ] ; then
echo "The amanda configuration MUST be the first argument to this script"
echo "And the TAPENUM must be the 2nd argument to this script"
exit 1
else
CONFIGNAME=$1
CONFPATH=$BASE_CONFPATH/${CONFIGNAME}
TAPENUM=$2
DATALOC="slot"$TAPENUM
echo "Line 24 DATALOC="$DATALOC
fi
# this is an attempt to derive the tape number for a rewrite of the archiving now done
# Change this hard coded path to whatever you are using for this particular cache, it does NOT
# go into your disklist!
#temp stuff these two
if [ $DUMMY -eq 1 ] ; then
echo "$0 started, mydir=$MYDIR config=$1 CONFPATH=${CONFPATH}"
fi
if [[ -d $MYDIR'/config-bak' ]] ; then
cd $MYDIR/config-bak
else
echo "ERROR: the data dir $MYDIR/config-bak does not exist"
exit 2
fi
# now get the final tape from the tapelist
# likewise, change this to your real path
TAPELIST=$CONFPATH/tapelist
#temp stuff
echo "Tapelist file to be used TAPELIST=${TAPELIST}"
# strip the tapename and tapenumber out of the string
if [ -f $TAPELIST ] ; then
read TAPEDATE TAPENAME TAPESTATUS junk < $TAPELIST
TAPENUM=${TAPENAME##*-}
echo "tapename=$TAPENAME, tapenum=$TAPENUM"
fi
# keep track of the passes on the tapes, sort of.
# first, check to see if old style name exists and use it if it does
#if [ -f dd.report.$TAPENUM ] ; then
# read TAPEUSECOUNT < dd.report.$TAPENUM
# echo "Line 57 Tape use count obtained from dd.report.$TAPENUM = $TAPEUSECOUNT"
#fi
# now, if that bombed
#if [[ $TAPEUSECOUNT -eq "" ]] ; then
echo "getting tape usage count from dd.report.$TAPENAME"
if [ -f dd.report.$TAPENAME ] ; then
read TAPEUSECOUNT < dd.report.$TAPENAME
echo "Line 71 Tape use count from dd.report.$TAPENAME = $TAPEUSECOUNT"
fi
echo "line 74 Tapeusecount is now="$TAPEUSECOUNT
# and if it still bombed out, reset the friggin count
if [[ $TAPEUSECOUNT -eq "" ]] ; then
TAPEUSECOUNT=0
fi
# increment it
TAPEUSECOUNT=$(( TAPEUSECOUNT + 1 ))
#some troubelshooting echo's
echo "This tape "$TAPENAME" has now been used $TAPEUSECOUNT times"
echo "INDICE_PATH=$INDICE_PATH, CONFPATH=$CONFPATH DATALOC=$DATALOC"
# now get rid of the data that was generated the last time this tape was used.
rm -f configuration.tar.$TAPENAME indices.tar.$TAPENAME dd.report.$TAPENAME
# this line can leave once all the stuff has been renamed else the tapenums are stuck in place!
rm -f configuration.tar.$TAPENUM indices.tar.$TAPENUM dd.report.$TAPENUM
# Lets see if its safe to write to the tape by generating a new dd.report.$TAPENAME
# and, so it gets into the email
touch dd.report.$TAPENAME
# Now set the tape use count as the first line of this refreshed file
echo "$TAPEUSECOUNT" > dd.report.$TAPENAME
# and report the invocation name used to line 2 (wrong slot# passed)
echo "This script is being run as " $0 $* | tee -a dd.report.$TAPENAME
# and report the tapename being used this time on line 3
echo "using tape $TAPENAME" | tee -a dd.report.$TAPENAME
# if using real tapes AND chg-scsi, enable this next line
# Is this needed ?
# mt -f /dev/nst0 tell | tee -a dd.report.$TAPENAME
# if that worked, we should have a current one
if [ ! -s "dd.report."$TAPENAME ] ; then # -s for exists AND is non-zero length
echo "dd.report."$TAPENAME" was not created or written to"
exit 3
fi
echo "tar up the indices and configs directories as they exist now." >> dd.report.$TAPENAME
echo "In bak-indices-configs DUMMY="$DUMMY >> dd.report.$TAPENAME
if [[ $DUMMY -ne 1 ]] ; then
# tar -cpsf indices.tar.${TAPENAME} $INDICE_PATH 2>&1 >> dd.report.$TAPENAME
tar -cpf indices.tar.${TAPENAME} $INDICE_PATH 2>&1 >> dd.report.$TAPENAME
# tar -cpsf configuration.tar.${TAPENAME} $CONFPATH 2>&1 >> dd.report.$TAPENAME
tar -cpf configuration.tar.${TAPENAME} $CONFPATH 2>&1 >> dd.report.$TAPENAME
else
tar -cpsf indices.tar.${TAPENAME} $INDICE_PATH
tar -cpsf configuration.tar.${TAPENAME} $CONFPATH
fi
# We've now generated the up2date indices & configs
# Figure out if the tape was written to using amstatus
PARTS_WRITTEN=`${AM_SBIN_DIR}/amstatus $CONFIGNAME | grep taped | awk -F: '{print $2}' | awk '{print $1}'`
# Ok, then lets make it part of the dd.report record
echo "Parts written = $PARTS_WRITTEN >> dd.report.$TAPENAME"
if [ $PARTS_WRITTEN -gt 0 ]; then
if [ $DUMMY -eq 1 ] ; then
echo "DUMMY="$DUMMY" indices.tar."$TAPENAME" not written"
echo "DUMMY="$DUMMY" configuration.tar."$TAPENAME" not written"
exit 7 # we ought to replace this with a defined error code just to be neat.
else
if [ $VTAPES -eq 1 ] ; then
echo "Using a vtape, so copying the indices & configuration to the vtape data dir."
# Amanda has stopped using the link 'data' to point to the correct tape, breaking this script
# rm -f ${VTAPE_LOCATION}/data/indices.tar # replace outdated file
rm -f ${VTAPE_LOCATION}/$DATALOC/indices.tar # replace outdated file
# cp indices.tar.$TAPENAME ${VTAPE_LOCATION}/data/indices.tar
cp indices.tar.$TAPENAME ${VTAPE_LOCATION}/$DATALOC/indices.tar
echo "Here are the contents of indices.tar" >> dd.report.$TAPENAME
# tar tf ${VTAPE_LOCATION}/data/indices.tar >> dd.report.$TAPENAME
tar tf ${VTAPE_LOCATION}/$DATALOC/indices.tar >> dd.report.$TAPENAME
# rm -f ${VTAPE_LOCATION}/data/configuration.tar # replace outdated file
rm -f ${VTAPE_LOCATION}/$DATALOC/configuration.tar # replace outdated file
# cp configuration.tar.$TAPENAME ${VTAPE_LOCATION}/data/configuration.tar
cp configuration.tar.$TAPENAME ${VTAPE_LOCATION}/$DATALOC/configuration.tar
echo "And here is the contents of configuration.tar" >> dd.report.$TAPENAME
# tar tf ${VTAPE_LOCATION}/data/configuration.tar >> dd.report.$TAPENAME
tar tf ${VTAPE_LOCATION}/$DATALOC/configuration.tar >> dd.report.$TAPENAME
else
#Real tape
echo "Using a real tape, so writing the indices & configuration using dd"
# once for the file & once for value
# due to the possibility of a double eof being written by amanda
mt -f /dev/nst0 seod
dd if=indices.tar.$TAPENAME of=/dev/nst0 bs=32768 conv=sync 2>&1 >> dd.report.$TAPENAME
if [ $? -ne 0 ] ; then
echo "dd command failed. Amanda configuration & indices not written to tape ! (tape
full?)"
exit 4
fi
echo "Here is the contents of indices.tar.$TAPENAME"
tar tf indices.tar.$TAPENAME >> dd.report.$TAPENAME
dd if=configuration.tar.$TAPENAME of=/dev/nst0 bs=32768 conv=sync 2>&1 >> dd.report.$TAPENAME
if [ $? -ne 0 ] ; then
echo "dd command failed. Amanda configuration not written to tape ! (tape full?)"
exit 5
fi
echo "And here is the contents of configuration.tar.$TAPENAME"
tar tf configuration.tar.$TAPENAME >> dd.report.$TAPENAME
mt -f /dev/nst0 weof 1
echo "Rewinding tape"
mt -f /dev/nst0 rewind
fi
fi
else
echo $TAPENAME" was not written to as amstatus $CONFIGNAME reported it had not written any partitions to tape." |
tee -a dd.report.$TAPENAME
exit 6
fi
# with the above data, we can try to troubleshoot if it fubars.
echo "The script "$0" is finished." | tee -a dd.report.$TAPENAME
echo "Amandatapes usage="`df /amandatapes` |tee -a dd.report.$TAPENAME
# The file /amanda/bak-configs/dd.report.$TAPENAME might have enough info
# to allow troubleshooting if the proceedure blows up someplace. One can
# always add more data outputs in the problem area... :-)
> > amanda uses has been updated in the last year or so and has broken
> > amanda somehow.
>
> Off hand, I am thinking there is a bug in the amstatus Perl code which
> gets triggered when you have a two-tape run, and then also a fix
> needed in your shell script so that PARTS_WRITTEN is always set
> correctly and the script can properly deal with a two-tape run.
>
> (Because you only see the error for the double-tape runs, I'm less
> inclinded to suspect a Perl upgrade is the issue [rather than a more
> general bug in amstatus, as it parsess the Amanda log file], but we
> may need to wait until the next time it happens before we can track it
> down.)
>
> Nathan
>
> ----------------------------------------------------------------------
>------ Nathan Stratton Treadway - [email protected] - Mid-Atlantic
> region Ray Ontko & Co. - Software consulting services -
> http://www.ontko.com/ GPG Key:
> http://www.ontko.com/~nathanst/gpg_key.txt ID: 1023D/ECFB6239 Key
> fingerprint = 6AD8 485E 20B9 5C71 231C 0C32 15F3 ADCD ECFB 6239
Copyright 2019 by Maurice E. Heskett
Cheers, Gene Heskett
--
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
If we desire respect for the law, we must first make the law respectable.
- Louis D. Brandeis
Genes Web page <http://geneslinuxbox.net:6309/gene>