Uhm, I see your point. Indeed now if I search for a not existing file with xrdcp -d 3 root://xrootd-redic.pi.infn.it//store/test/xrootd/T2_ES_IFCA/store/mc/SAM/GenericTTbar/GEN-SIM-RECO/CMSSW_5_3_1_START53_V5-v1/0013/CE4D66EB-5AAE-E111-96D6-003048D37524.rootf . I get redirected to cern (global redir) via 140709 07:49:38 6036 Xrd: HandleServerError: Received redirection to [cms-xrd-global.cern.ch:1094]. Token=[]]. Opaque=[tried=+1213xrootd.ba1213xrootd-redic.pi.infn.it]. so opaque seems a mangled version of BOTH cmsd. BUT: If I go back to the old way (connecting each redirector to ITS OWN cmsd, not to the DNS, I mean going back to all.manager xrootd-redic.pi.infn.it 1213 instead of all.manager xrootd-cms.infn.it+ 1213 ) I get 140709 07:52:44 6635 Xrd: HandleServerError: Received redirection to [cms-xrd-global.cern.ch:1094]. Token=[]]. Opaque=[tried=+1213xrootd-redic.pi.infn.it]. So in this case going up and down would work, right? because when coming down it will try xrootd.ba.infn.it:1213 since it is not in the opaque it would be like having 2 EU redirectors which do not speak to each other, and connect to the global one. In this case, an unbalance in their servers (even transient) would be cured by the global redirector. Exactly why this is not supposed to be ok? Looking at the first email, for a really non existing file, it can try once per redirector, but I would prefer to pay this price here ... ciao ciao tom --- Reply to this email directly or view it on GitHub: https://github.com/xrootd/xrootd/issues/124#issuecomment-48432831 ######################################################################## Use REPLY-ALL to reply to list To unsubscribe from the XROOTD-DEV list, click the following link: https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-DEV&A=1