Print

---

Doug,

The reason is that you misspelled the global redirector's hostname, It is 
not "grd....." but "glrd....".

Andy

-----Original Message----- 
From: Yang, Wei
Sent: Wednesday, August 31, 2011 12:16 PM
To: Doug Benjamin
Cc: xrootd-l ; Hironori Ito ; Ofer Rind
Subject: Re: Odd behavior with a meta manager

Doug,  if you have a single proxy, why the query returns two machines?

regards,
Wei Yang  |  [log in to unmask]  |  650-926-3338(O)


On Aug 31, 2011, at 6:27 AM, Doug Benjamin wrote:

> Dear XRoot folks,
>
>  I am seeing some odd behavior with a meta manager.   The configuration is 
> such that
> a simple proxy server is used to connect to a meta manager.  I am testing 
> the existence of
> a directory (which I know exists) on a data server and local redirector 
> that are dual NIC's
> and are only a private network. All of the traffic between the local 
> redirector and data server
> is through the private network only.
>
> I am testing for the directory through the meta manager from two different 
> machines. 1) on the
> same network as the local redirector and one offsite.   Here is my testing 
> squence:
>
> 1) From Machine offsite (atl010.phy.duke.edu)
>
> [benjamin@atl010 ~]$ xrd grd.usatlas.org:1094 locateall 
> /atlas/ANLASC_TEST/mc10_7TeV.105802.JF17_pythia_jet_filter.merge.NTUP_SUSY.e577_s933_s946_r1831_r2040_p428_tid281325_00
> Not connected to any server.
>
> 2) shortly afterwards from machine on the same network as the local 
> redirector and data server (onsite) (ascint0y.hep.anl.gov)
>
> [dbenjamin@ascint0y ~]$ date
> Wed Aug 31 07:37:40 CDT 2011
> [dbenjamin@ascint0y ~]$  xrd glrd.usatlas.org:1094 locateall 
> /atlas/ANLASC_TEST/mc10_7TeV.105802.JF17_pythia_jet_filter.merge.NTUP_SUSY.e577_s933_s946_r1831_r2040_p428_tid281325_00
>
>
> ------------- Location #1
> InfoType: kXrdcLocDataServer
> CanWrite: false
> Location: '130.202.173.4:1094'
>
>
> ------------- Location #2
> InfoType: kXrdcLocDataServer
> CanWrite: false
> Location: '130.202.173.42:41094'
>
>
> 3) Immediately aftwards repeat the query on the offsite machine (note the 
> machines are in two different time zones)
> [benjamin@atl010 ~]$ date
> Wed Aug 31 08:38:04 EDT 2011
> [benjamin@atl010 ~]$ xrd glrd.usatlas.org:1094 locateall 
> /atlas/ANLASC_TEST/mc10_7TeV.105802.JF17_pythia_jet_filter.merge.NTUP_SUSY.e577_s933_s946_r1831_r2040_p428_tid281325_00
>
>
> ------------- Location #1
> InfoType: kXrdcLocDataServer
> CanWrite: false
> Location: '130.202.173.4:1094'
>
>
> ------------- Location #2
> InfoType: kXrdcLocDataServer
> CanWrite: false
> Location: '130.202.173.42:41094'
>
> Both machines (atl010.phy.duke.edu and ascint0y.phy.duke.edu) have the 
> same time to within 2 seconds of each other. (except for the 1 hour time 
> zone difference)
>
> Any Idea why the first time the remote machine did not get any answer 
> back?
>
> Finally, how long does it talk for a meta manager to drop a data server 
> (or proxy)  from its list automatically. (The second proxy 
> (130.202.173.42:41094)  was turned off ~6 hours
> ago.
>
> Thanks,
>
> Doug Benjamin
>
>