buffindexed.c:1525 Can't malloc 273997900 bytes: Not enough space

Gautam Das gautam at ufl.edu
Fri May 17 18:30:43 UTC 2002


Joe St. Sauver of uoregon.edu helped me out the malloc error. I was 
hitting a default AIX data segment size limit of 256 MB. I recompiled with 
 -Wl,-bmaxdata:0x20000000 and that fixed the malloc problem.

However I still have buffindexed errors, as I see lots and lots of "could
not open overview for ..." errors. So I am getting ready to move back to
tradindexed. I am aware of the performance problems with tradindex, but do
I have another choice?  The general feel I got from several people is to 
stay with tradindexed, as buffindexed and ovdb are not yet ready for primetime.
Woudl you argue otherwise?

-- 
Gautam 

On Sat, 18 May 2002, Katsuhiro Kondou wrote:

> 
> In article <Pine.LNX.4.44.0205101504160.1544-100000 at pluto.nerdc.ufl.edu>,
> 	Gautam Das <gautam at ufl.edu> wrote;
> 
> } expireover is exiting after running for about 6 hours instead of taking 
> } the usual 12 to 14 hours, and in the news.daily report I am seeing the 
> } following error:
> } 
> } buffindexed.c:1525 Can't malloc 273997900 bytes: Not enough space
> 
> If you have huge number of articles in some newsgroup,
> it may happen.  But,
> 
> } expireover[34302]: buffindexed: could not open overview for 
> } 'alt.autos.ford'
> } expireover[34302]: buffindexed: ovgroupmmap ovbuff is null(ovindex is 
> } 12599,  ovblock is 925972024
> 
> may affect the problem.
> 
> } Is my overview data is inconsistent. Whats wrong? 
> 
> It looks like somewhat corrupted.  Did your server crashed?
> 
> } Shall I consider moving overview to tradindexed?
> 
> The problem should be gone if you move to tradindexed.
> But if you want to gain performance, tradindexed is not
> so fast in high volumed environment.
> 




More information about the inn-workers mailing list