 |
 |
Hi, XrootD experts. I have been working on AAA file-opening scale tests, and, in the process, have been examining error messages from XrootD. Unfortunately, in some cases these messages are not helpful and may not even be accurate. A generic error message is the following: Xrd: CheckErrorStatus: Server [xrootd.unl.edu:1094] declared: No servers are available to read the file.(error code: 3011) Xrd: Open: Error opening the file. It can mean that a site is overwhelmed with too many requests, or that a requested file doesn't exist. It is a very common error that provides little clue as to what is actually wrong. Here's a message that seems to indicate a file doesn't exist: Xrd: CheckErrorStatus: Server [xrootd.rcac.purdue.edu:1094] declared: Unable to open /store/test/xrootd/T2_US_Purdue/store/data/Run2012A/PhotonHad/AOD/13Jul2012-v1/00000/7EC77E00-66CE-E111-A2F8-00215E93C8B0.root; no such file or directory(error code: 3011) In contrast, at the same time this message was issued, the XrootD server reported that it successfully opened the same file. In this case, this error message is simply false, since this file does, in fact, exist at the site. At a site where XrootD is running well, I have observed that file-opening errors are entirely due to attempts to open non-existent files. Unfortunately, if the request is made through the Nebraska XrootD redirector, the error message in this case will be the generic "No servers available" message and not the more specific "no such file" message. It would be helpful if the XrootD error messages could be made more accurate and meaningful. In particular, the server should always issue a distinctive "no such file" message for non-existent files and never do so for files that do exist. Secondly, additional information in the error message would be helpful. Perhaps the message could include more details about the script or module that encountered the error. Also, the messages have error codes, but almost all of them have the value 3011, even though the messages are different. I'm not sure what numbering scheme there is for the errors, but it might be helpful to give each message its own code. If anyone would like further information about my observations of XrootD errors, or if I could assist in any other way, please let me know. ---Carl Vuosalo UW-Madison ######################################################################## 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