CachedDataStore.pm line 92

[marsha]

I have cpanel backup running to a drive and now the resellers are not being backed up, but regular sites are, when I looked in the cpbackup folder I found tar files for all my sites and resellers plus folders for the resellers, and 0 byte files for them. The reseller tar files were 0 byte the others seem good I used one to restore a site and it worked find
The error I am getting is

[cpbackup] Started at Tue Jun 17 01:00:01 2008
pkgacct started.
pkgacct version 7.1 - archive version: 2 - running with uid 0
using time::hires for speedups
Copying Reseller Config...can't get exclusive lock on /backup//cpbackup/daily/thetopto/resellerconfig/my_reseller-limits.cache: No locks available at /scripts/cPScript/CachedDataStore.pm line 92
pkgacct started.
pkgacct version 7.1 - archive version: 2 - running with uid 0
using time::hires for speedups
Copying Reseller Config...can't get exclusive lock on /backup//cpbackup/daily/bonofi/resellerconfig/my_reseller-limits.cache: No locks available at /scripts/cPScript/CachedDataStore.pm line 92
pkgacct started.
pkgacct version 7.1 - archive version: 2 - running with uid 0
using time::hires for speedups

What am I missing?

[cpdan]

Hello,

This happens when something already has a system flock() on the given .cache file.

I'm looking into a solution now, would you be able to create a ticket so I can investigate the issue on a box that is exhibiting this behavior?

Just make it 'ATTN: Dan' and reference this forum thread

thanks

[cpdan]

I can't reproduce this, I'll definitley have to have a sytem that exhibits this problem as it appears to be a very low level flock() issue.

A good starting point woudl be the output of this command on said box:

perl -le 'require Config;print "$_ : $Config::Config{$_}" for qw(d_flock d_fcntl_can_lock d_lockf);'

[cpdan]

Also this may be helpful to know:

perl -le 'use POSIX;use Storable;print $];print $POSIX::VERSION;print $Storable::VERSION;'

[marsha]

This is what I got. I will open a ticket...


root@cor1web [~]# perl -le 'require Config;print "$_ : $Config::Config{$_}" for qw(d_flock d_fcntl_can_lock d_lockf);'
d_flock : define
d_fcntl_can_lock : define
d_lockf : define
root@cor1web [~]#


root@cor1web [~]# perl -le 'use POSIX;use Storable;print $];print $POSIX::VERSION;print $Storable::VERSION;'
5.008008
1.09
2.18
root@cor1web [~]#

[cpdan]

Thanks, the only differnece i note off hand is the Storable version. However no n of the changes appear, at first glance anyway, to effect the flock() calls:

http://search.cpan.org/diff?from=Sto...=Storable-2.18

I look forward to your ticket!

[cpdan]

/backup does not do flock(2)

root@cor1web [~]# touch /testroot@cor1web [~]# perl -e 'use Storable;open(my
$fh, "<", $ARGV[1]) or die "open: $!";print
"OPENED\n";alarm(5);flock($fh, 2) or die "flock:
$!";alarm(0);print "FLOCKED\n";alarm($ARGV[0]) if
$ARGV[0];Storable::lock_store({}, $ARGV[0]);print "STORED\n";' 4
/test
OPENED
FLOCKED
STORED
root@cor1web [~]# touch /backup/testroot@cor1web [~]# perl -e 'use
Storable;open(my $fh, "<", $ARGV[1]) or die "open:
$!";print "OPENED\n";alarm(5);flock($fh, 2) or die "flock:
$!";alarm(0);print "FLOCKED\n";alarm($ARGV[0]) if
$ARGV[0];Storable::lock_store({}, $ARGV[0]);print "STORED\n";' 4
/backup/test
OPENED
flock: No locks available at -e line 1.
root@cor1web [~]#

because /backups is an NFS mount

XXX.XXX.XXX.XXX:/XXX/backup /backup nfs defaults 0 3

[cpdan]

NFS problem worked around in rev 25608

[marsha]

Thank you very much...

I did the nfs to locate my backups on a machine here that is located in a 1947 bank vault thinking anti theif and fireproof... Insurance will replace machines but not data..:}

Marsha Black

[cpdan]

No problem, NFS is persnickety

[astopy]

I'm getting this same problem on the latest STABLE. Looking at the changelog, it looks like the workaround is only in EDGE?

This isn't a huge deal for me, as the only reseller accounts we have are staff/admin accounts, so it's not vital that they be backed up, but do you have an estimate on how long it will be until this hits STABLE?

[jmcole]

We are upgrading our production server to EDGE to get this fix, we'd really appreciate it if this was moved into Release or Current at minimum, thank you!

We have no choice as the reseller account on this particular server has over 20GB of data to backup to our new NFS mounted /backup. Can't risk running ftp/compressed backups considering the size of their account so we have no other choice it seems at this time!

I hate running production servers on EDGE, wish us luck.

[cpanelkenneth]

We are upgrading our production server to EDGE to get this fix, we'd really appreciate it if this was moved into Release or Current at minimum, thank you!

We have no choice as the reseller account on this particular server has over 20GB of data to backup to our new NFS mounted /backup. Can't risk running ftp/compressed backups considering the size of their account so we have no other choice it seems at this time!

I hate running production servers on EDGE, wish us luck.

CURRENT was published last night with this fix.