WIZARD@RITA.ACS.WASHINGTON.EDU (The Bandit "." "." "." "", on" "RITA) (02/12/88)
Several people have commented about the strange behaviour of DIRECTORY recently. Fred Zwarts writes: > But try the following commands when you have a cluster: > > $ SET DEFAULT SYS$SYSROOT:[000000] > $ DIRECTORY SYS$SYSDEVICE > > You will also see each file listed twice. Another netter mentioned that doing a DIRECTORY of a TU81+ device listed each file twice. Well, I found a similar problem, but it cannot be traced to a search list, since I don't use search lists and we don't (currently) cluster our VAXes. In fact, I have seen this particular bug at another stand-alone site. Here is the setup. My login procedure defines a rooted device which is configuration specific. All my other directory logicals are defined in terms of this "device." Thus, $ ASSIGN /TRANSLATION=(CONCEALED,TERMINAL) DUA2:[000000.] ROOT$DEVICE $ ASSIGN ROOT$DEVICE:[DH00.MACRO] $MAC One day I was trying to determine if the quota files were messed up again and decided to compute the actual usage for a disk. Therefore, I set my default "up" to the top level directory on the disk, effectively with the command $ SET DEFAULT ROOT$DEVICE:[000000] I then entered the following command $ DIRECTORY /SIZE=ALL /GRAND [...] and waited. And waited. And waited. And waited. etc... Finally, after 15 or so minutes, I aborted the command and changed it to $ DIRECTORY /SIZE=ALL /TOTAL [...] I discovered that DIRECTORY was screwing up badly. (Currently we are running VMS V4.5, but I saw this at least as early as VMS V4.2.) Here's how. Let MFD = Main File Directory, and UFD_n = User File Directory number "n". Directory shows me the contents of MFD as being in each UFD. Thus, it appears that [UFD_1] === [MFD], with only a name change. Now, I don't have any "small" disks that I feel like letting this go on forever, and maybe it would terminate after two complete passes, but my impression has always been that DIRECTORY gets into an infinite loop in this particular case. Comments? Anyone care to tell me if this is still the case in VMS V4.6 and/or VMS V4.7? Thanx, Derek Haining Academic Computing Services University of Washington Seattle, Washington (206) 543-5852 DEREK@UWARITA.BITNET -or- DEREK@RITA.ACS.WASHINGTON.EDU