Conference iosg::all-in-1_v30

1146.0. "RSF run failure and system crash" by KEMER::YEGEN (the HOTLINE dude) Thu Jul 30 1992 11:32

For ALL-IN-1 V2.4 we are using three disks first for ALL-IN-1 files, second 
for SHARE and third for SHARD. On the second disk OA$DAF_E.DAT has become
700000 blocks (and I had closed this area 2 weeks ago) and since there are 
ca. 200000 blocks are free on this disk RSF procedure fails (the part of the 
log file is shown below)

Yesterday I defined into LOGIN.COM of ALLIN1 account a line like 
DEFINE SYS$SCRATCH dev:dir  (which had more than 1400000) free
and the result was very interesting the first ALL-IN-1 disk was full and RSF
was running more then 6 hours and one of the nodes in the clusters was down
and the other node was very slow after deleting RSF all the housekeeping 
procedures begun to run and I could only solve the problem by rebooting
both nodes and after the reboot there were no housekeeping procedures in
any queu which were waiting for their next run.

Can anybody suggest me how to make the run RSF successfully.


Best regards,

Tansu YEGEN @TWI

==============================================================================
LOG FILE FOR AN UNSUCCESFULL RSF RUN
==============================================================================

%SMREORG-I-CEND "$1$DUA14:[OA$SHARD]OA$DAF_D.DAT" converted - was 26532 blocks,
 now 19114 blocks
%SMREORG-I-CSTART Converting "A1$SYS_DISK:[ALLIN1.SHARED_E]OA$DAF_E.DAT" 
using FDL file "OA$LIB:SDAF.FDL"
%CONVERT-F-EXTN_ERR, unable to extend output file
-RMS-F-FUL, device full (insufficient space for allocation)
%SMREORG-E-CFAIL Convert failed for "A1$SYS_DISK:[ALLIN1.SHARED_E]OA$DAF_E.DAT"
%NONAME-E-NOMSG, Message number 00000002
%SMREORG-E_UNEXEC Unexpected error condition
-NONAME-E-NOMSG, Message number 00000002
%NONAME-E-NOMSG, Message number 00000002
%SMJACKET-I Invoking common END_BATCH procedure
%END_BATCH-I Performing End batch job procedures for "RSF"

    I think your only option here is to edit the command procedure to do
    the convert onto another disc and then copy the new version of the
    fileback again afterwards.
    
    The System Manager of our production system here beat me up about this
    only last week!
    
    Sorry,
    
    Graham