CondorMonitor improperly handles security changes in case of exeptions
I just found out that the CondorStatus class does not properly clean up ins internal security changes in case of an exception.
Other classes in the CondorMonitor library likely have the same problem.
#1 Updated by Igor Sfiligoi over 5 years ago
Noticed this as a side effect of the glidein_off bug;
on retry, CondorStatus will complain like so:
Traceback (most recent call last): File "glidein_off", line 301, in <module> sys.exit(main(sys.argv)) File "glidein_off", line 240, in main addrs,failures = lookup_addrs(complete_constr) File "./glidein_off", line 154, in lookup_addrs data = sts.fetch(master_constr,[('MyAddress','s')]) File "lib/condorMonitor.py", line 342, in fetch return QueryExe.fetch(self, constraint=constraint, format_list=format_list) File "lib/condorMonitor.py", line 266, in fetch self.security_obj.save_state() File "lib/condorSecurity.py", line 131, in save_state raise RuntimeError,"There is already a saved state! Restore that first." RuntimeError: There is already a saved state! Restore that first.