Jose Maria Blasco <JMBLASCO@DEARN>
Mon, 29 Aug 88 15:16:15 EST
|
Since we are going to VM/XA SP1 in our mainframe (we are currently
running XA SF2), I decided to install CMS 5.5 under the VM/SP guest to
see how it worked. Well, MAILER didn't want to start, other utilities as
FLIST, BROWSE or LOOK produced random abends, and so on, but LISTSERV
seemed to work. Fine. So I left it running during the night to see what
happened. This morning I had about 6 error mails from LISTSERV telling me
that there had been a rexx error. By inspecting the console log I found
that this was LSVLUPD which crashed (machine resources exhausted) when
processing X-LUPD commands. Then I made the following experiment: I
isolated the piece of code that was crashing (it is copied below) and I
run it. These are the results (for a GLOBLIST FILE with 1150 entries):
* On CMS 5.0, it runs fine with a 2M machine.
* On CMS 5.5, it crashes with a 2M machine (after processing 450
entries).
* With 4 megs, it crashes after processing 760 entries.
* With 8 megs it does run, *but* on listserv it still crashes, probably
because LISTSERV is already using part of the memory.
Well now this was done on LISTSERV@DBNGMD12, where nobody is using the
DATABASE facilities, which use similar catenation strategies. I don't
know if there will be a fix for that or if this is part of the improved
support for storage management that IBM has so proudly announced...
Jose Maria
--------------------------------------------------------------
/**/
list = ''
'LSVFILER GLOBLIST FILE (STEM L.'
Say l.0
Do i = 1 for l.0
Parse var l.i id nodeid rest
list = list id'@'nodeid
info.nodeid.id = rest
End
Drop l.
|
|
|