newman@venus.ycc.yale.edu (01/17/90)
Greetings: A bit ago someone posted a message describing a problem adding large batches under NEWS 5.9B. The log indicated what appeared to be an access violation when creating the text file. I had the same problem but the reason was none of the reasons proposed.... The symtoms were that each entry in the batch created an entry in the JUNK group but the text of the message was missing from the JUNK directory. This happened mysteriously in the middle of a large batch add. Further investigation revealed that attempting to create ANY file of ANY type on the NEWS disk volume generated the VMS error message RMS-F-HEADERFUL. As the messages manual states, this indicated that the file header map area is/was full. Their suggested USER ACTION was to backup the volume to tape and restore it. How this might help is a mystery to me. I did it, it helped, but I'm still a little disconcerted. The disk in question had PLENTY of free block on it. There were just under 50000 files on the volume and the estimated maximum number of header entries should be somewhere around twice that figure. The disk is an RA81 initialized with the default parameters. Can anyone: 1) Explain how restoring a volume from disk solves this problem... 2) Come up with a series of reasonable precautions to avoid the problem in the future. I have some general ideas involving re-initializing the volume with different /CLUSTER and /MAXIMUM_FILES parameters but because of the time involved and the lack of information I have, I am appealing to the NET for suggestions and/or war stories along a similar track.... -Andy Newman Software Systems Programmer Yale University....
glassmann@ccavax.camb.com (01/19/90)
> VMS error message RMS-F-HEADERFUL. As the messages manual states, > this indicated that the file header map area is/was full. Their > suggested USER ACTION was to backup the volume to tape and restore > it. How this might help is a mystery to me. I did it, it helped, > but I'm still a little disconcerted. > It seems as if some of your files were fragmented enough that they needed multiple headers. DUMP/HEADER will show you how many headers a file has. -- Lenny Glassmann lenny@ccavax.camb.com ...uunet!ccavax!lenny
gih900@CSC.DNET (Geoff Huston) (01/22/90)
>The symtoms were that each entry in the batch created an entry in the >JUNK group but the text of the message was missing from the JUNK >directory. This happened mysteriously in the middle of a large >batch add. Further investigation revealed that attempting to >create ANY file of ANY type on the NEWS disk volume generated the >VMS error message RMS-F-HEADERFUL. As the messages manual states, >this indicated that the file header map area is/was full. Their >suggested USER ACTION was to backup the volume to tape and restore >it. How this might help is a mystery to me. I did it, it helped, >but I'm still a little disconcerted. > >The disk in question had PLENTY of free block on it. There were just >under 50000 files on the volume and the estimated maximum number of >header entries should be somewhere around twice that figure. The >disk is an RA81 initialized with the default parameters. > >Can anyone: > > 1) Explain how restoring a volume from disk solves this problem... > > 2) Come up with a series of reasonable precautions to avoid the > problem in the future. > >I have some general ideas involving re-initializing the volume with >different /CLUSTER and /MAXIMUM_FILES parameters but because of the >time involved and the lack of information I have, I am appealing to >the NET for suggestions and/or war stories along a similar track.... The MAXIMUM_FILES parameter is the one too look at - NEWS creates LOTS of small files which fills up the allocated header space in INDEXF.SYS. You can find out more about this by posting to comp.os.vms Geoff Huston -Andy Newman Software Systems Programmer Yale University....