7hrs and rising...
Richard Michael Todd
rmtodd at mailhost.ecn.ou.edu
Fri Nov 5 23:33:50 UTC 1999
In message <Pine.BSF.4.10.9911051727560.2296-100000 at thelab.hub.org>you write:
>
>Am running 'makedbz -o' on a 654Meg history file, and its still going
>strong?? I'm on a PIII-500Mhz with 768Meg of RAM...shouldn't it run just
>slightly faster then that? :)
Yeah. On a PPro 200 with 128M and a 250M or so history file, makedbz
-s 2808378 (the history file length) takes 58 seconds.
As a later correspondent whose name I can't get to easily while I'm writing
this message writes, the most likely cause of this long time is not having
the -s parameter set right. *However*, closer inspection reveals that
1) the -o parameter, according to the man page, is supposed to use
the size of the old history file to autosize the new one, and yet
2) the actual code for makedbz says:
case 'o':
Overwrite = TRUE;
IgnoreOld = TRUE;
break;
IgnoreOld does exactly what you might guess, making it ignore the old history
dbz size. Hmm. From the looks of the code, it looks like the -o option has
been blithely ignoring the old DBZ size since well before INN2.2. Try the
enclosed patch which seems to make -o work the way the man page says.
Index: expire/makedbz.c
===================================================================
RCS file: /home/rmtodd/TempCVS/Temp-CVS-Repository/inn-current/expire/makedbz.c,v
retrieving revision 1.5
diff -u -r1.5 makedbz.c
--- makedbz.c 1999/07/04 05:56:45 1.5
+++ makedbz.c 1999/11/05 23:25:33
@@ -97,7 +97,11 @@
}
else {
temp[0] = '\0';
- RemoveDBZFiles(TextFile);
+ /*
+ ** only do removedbz files if a. we're using something besides
+ ** $pathdb/history, or b. we're ignoring the old db.
+ */
+ if (!EQ(TextFile, HISTORY) || IgnoreOld) RemoveDBZFiles(TextFile);
p = TextFile;
}
@@ -230,7 +234,6 @@
break;
case 'o':
Overwrite = TRUE;
- IgnoreOld = TRUE;
break;
case 'i':
IgnoreOld = TRUE;
More information about the inn-workers
mailing list