 |
 |
Hi Paul, both CNS and SSI (which uses CNS) are supported but we don't find a use case right now. Their functionality are relatively separated from the rest so I don't expect other development will introduce serious bug to those two. Of course, this is just a guess. if you data servers are extremely busy, xrootdfs will be slow unless you use CNS and only give dir/file stat() info from CNS (this is quite useful and reliable but not 100% as the CNS updates are not atomic). I had this problem at SLAC before, and we can address the problem not by going back to CNS, but by reconfigure the layout of disks on data server to speed up things, and completely turn off disk and filesystem read ahead. I think the CNS is left there to cover new types of storage/hardware/configuration that we don't have an immediate good solution, However finding out why xrootdfs is slow (and address the backend disk performance if needed) is probably the best way to do. regards, Wei Yang | [log in to unmask] | 650-926-3338(O) On Jul 3, 2012, at 11:52 AM, Paul T. Keener wrote: > Wei, > > Thanks. > > I was trying to use it to solve problems I was having with apparent > timeouts opening files when the data server was very busy as well as > improve dreadful latency problems getting directory listings on xrootdfs. > > If CNS isn't the way to improve the situation for these problems, what is? > > Is SSI obviated as well? > > Thanks again. > > -paul > > > > On 07/03/2012 02:29 PM, Yang, Wei wrote: >> I guess the answer is yes. it is still supported though we don't have a use case right now. >> >> regards, >> Wei Yang | [log in to unmask] | 650-926-3338(O) >> >> >> On Jul 3, 2012, at 10:03 AM, Paul T. Keener wrote: >> >>> Is CNS still supported? >>> >>> I have tried to find information on getting it set up correctly, but the >>> existing documentation is incomplete and I have received no response to >>> my questions. >>> >>> Thanks. >>> >>> Paul T. Keener >>> Department of Physics and Astronomy >>> University of Pennsylvania >>> >>> ######################################################################## >>> Use REPLY-ALL to reply to list >>> >>> To unsubscribe from the XROOTD-L list, click the following link: >>> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1 >> >> ######################################################################## >> Use REPLY-ALL to reply to list >> >> To unsubscribe from the XROOTD-L list, click the following link: >> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1 ######################################################################## Use REPLY-ALL to reply to list To unsubscribe from the XROOTD-L list, click the following link: https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1