 |
 |
Hi All, We would like to start doing some more quantative testing of xrootd here in the ARDA project. For this I looked over the monitoring capabilities in xrootd and since I have some experience with MonALISA I wrote a module that receives and decodes the UDP packets sent by xrootd. However, I have some problems interpreting the data that I receive and after a discussion with Jacek about this, I was directed to the mailing list. First of all, here is my setup: I have 2 machines: - monalisa.cern.ch - here is the dataserver(xrootd 20050417-0431), having in configuration: xrootd.monitor all flush 1m window 1m dest files io info user pcardaab.cern.ch:8901 - pcardaab.cern.ch - here I run my MonALISA module that receives the data. To test the system, I run: > xrdcp -DIDebugLevel 2 root://monalisa//tmp/file-4.ml . Here is the list of encountered problems. Also please check the attached output generated by the module. 1) Sending the I/O trace packet after the specified period of time doesn't seem to work. From what I've seen in the documentation, the "flush" parameter for xrootd.monitor from the dataserver's configuration file should do this. You can see in the generated output that I have only one i/o trace datagram for the whole session, althouth it lasted more than 1 minute. 2) "end of last window" == "start of this window" for each two consecutive window entries from the i/o trace. You can grep the attached file for WINDOW to see this. Is this correct? 3) Packet lenght is reported as being 0, for each of the packets (either map or trace). 4) I think that the number of bytes actually sent after a READ request could be more useful than the number of the requested bytes, in order to have a better precision when computing the total number of bytes sent to the clients by a xrootd dataserver. Thanks for your help, Catalin.