Red Hat Bugzilla – Bug 623238
jobserver: problems reconstructing history job ads
Last modified: 2010-10-12 16:28:02 EDT
Further testing/investigation required but it appears if a history entry has an ad that includes certain characters (ie., *) then the ClassAd ctor with a "***" delimiter gets confused and throws an error.
This was observed on mrg31 where some jobs were being sent in with Environment like "u'Environment': "LS_COLORS=no=00:fi=00:di=00;34:ln=00;36:pi=40;33:so=00;35:bd=40;33;01:cd=40;33;01:or=01;05;37;41:mi=01;05;37;41:ex=00;32:*.cmd=00;32:*.exe=00;32:*.com=00;32:*.btm=00;32:<snip>"
ClassAd ctor throws an error saying "bad expr..." but it is not always directly traceable to individual ads which suggests a cumulative parsing error across ads in the file potentially.
It's not clear that this has anything to do with wildcards. More likely that the offsets in the history entries managed by the job server and the indices are somehow "driftng". condor_history -l of randomly selected clusters looks ok.
Modified with FH sha 634e1cd
Significant reworking of how summaries are collected from Live and Historical Jobs so will reassess once condor is updated on mrg27