We are running a few zones on solaris and have a clientbucket problem
on a few of them. Occassionaly we have clientbuckets with a size of
30G and more. Our theory is that we have such a large amount of
data in the clientbucket directory, because of the fact that the
mounted share was modified while it was mounted. Then puppet seems to
remount and to rescue the old state to the directory clientbucket?!
We have investigated all of our manifests and - in my opinion - the
following part is interesting in this case:
[...]
"tests.executed" : {
$resultpath = "/mnt/systemtest/$product_release/Results/FunctionalTest/NightlyTest/$test_suite/$test_scenario/archive"
$sopbase = "/opt/SOPware"
#
# FIXME: Additional logs (e.g. from AppServers)
# must be included!
#
zip {
"/tmp/testlogs.zip" :
source => ["$BaseDir/shared/log ",
"$BaseDir/shared/conf ",
"$BaseDir/shared/temp ",
"$BaseDir/participants/*/conf ",
"$BaseDir/participants/*/temp ",
"$BaseDir/participants/*/log ",
"$sopbase/install.log ",
"$sopbase/*/log/* ",
"$sopbase/*/logs/* ",
"$sopbase/*/sopware.params ",
"/tmp/*log "
# joram
# mq
# etc
],
vorher => Exec["copy-logzip"]
}
[...]
exec {
"prov-testresdir" :
command => "mkdir -p $resultpath/$testrun-$testbedname",
cwd => "/tmp",
path => "/usr/bin:/usr/local/bin:/bin:/sbin:/usr/sbin",
creates => "$resultpath/$testrun-$testbedname"
}
[...]
exec {
"copy-logzip" :
command => "cp /tmp/testlogs.zip $resultpath/$testrun-$testbedname/",
cwd => "/tmp",
path => "/usr/bin:/usr/local/bin:/bin:/sbin:/usr/sbin",
creates => "$resultpath/$testrun-$testbedname/testlogs.zip"
}
[...]
Assuming 10 zones running on one container, we must check frequently all client buckets to ensure our production.
This ticket is send on behalf of moett, who does not have sufficient rights to create tickets.
