\item Purge all data on the server with \path{glite-lb-purge}
\item Log two sets of jobs, separated with delay of at least 60s so
that the sets can be distinguished from each other.
-\item \label{purgel}
+\item \label{purge1}
Using \code{edg\_wll\_JobLog} retrieve events of all the jobs
\item \label{purge2}
Purge the first set of jobs (by specifying appropriate timestamp),
letting the server dump the purged events.
\item \label{purge3} Purge the other set of jobs, also dumping the events.
\item \label{purge4} Run purge once more.
+\item Check if purged jobs turned into zombies.
+\item In addition, check if a \emph{cron} task exists to run the \emph{purge} operation regularly and that it logs its output correctly.
\end{enumerate}
\how\ \ctblb{lb-test-purge.pl}
-\note\ The test includes artificial delays. Takes approx. 2\,minutes to finish.
+\note\ The test includes artificial delays. Takes approx. 3.5\,minutes to finish.
-\note\ This test is destructive to your data. You need to run it with the \texttt{--i-want-to-purge} option to confirm your intention. Also, you need to provide the \LB server \texttt{address:port} explicitly as an argument to rule out any confusion.
+\note\ This test is destructive to your data. You need to run it with the \texttt{-{}-i-want-to-purge} option to confirm your intention. Also, you need to provide the \LB server \texttt{address:port} explicitly as an argument to rule out any confusion.
-\result\ The data dumped in steps \ref{purge1}, \ref{purge2} should be the
-same as retrieved in~\ref{purgel}. The final purge invocation should
+\result\ Data dumped in steps \ref{purge1} and \ref{purge2} should be the
+same as retrieved in~\ref{purge1}. The final purge invocation should
do nothing (i.e. nothing was left in the database).
% test_purge
git://scientific.zcu.cz / jra1mw.git / commitdiff