Syslog Entry

Vernon Schryver
Sun Mar 31 17:01:17 UTC 2002

> From: Phil White <>

> Mar 31 01:23:15 soth dccd[3401]: initialize flood stream to
> Mar 31 01:23:15 soth dccd[3401]: 1.0.52 listening to port 6277 with /var/lib/dcc
> Mar 31 01:23:17 soth dccd[3401]: ask,6277 to rewind flood
> Mar 31 01:28:12 soth dccd[3401]: iflod status fail=1,ok=0 to no free hash table entry; 21504 of 21504 used: 24603 received  16515 accepted  0 duplicate  8088 stale  0 bad whitelist  0 not deleted  0 unknown ID

What is with the \xa0 characters instead of blanks (\x20) before the
zeros in that line?  Does the tool used to capture the log entry
convert some blanks to blanks-with-parity?  Is there something screwy
with the printf() library on the target platform?  I don't think it's
a bug in the DCC source, but am I wrong?

> Mar 31 01:28:16 soth dccd[3401]: add host_id claim: no free hash table entry; 21504 of 21504 used
> Mar 31 01:28:16 soth dccd[3411]: starting `dbclean -PqR -i 1022 -L info,MAIL.notice -L error,MAIL.err`
> Mar 31 01:28:16 soth dbclean[3411]: 1.0.52 repairing dcc_db
> Mar 31 01:28:17 soth dccd[3401]: database cleaning begun
> Question: Are lines 4,5,7 anything to worry about?
> Can I have an english interpretation of this please. I read it ATM as
> "Can't flod. I havn't got any spare room to store checksums. Am starting database repair"
> Except that I issued a
> DBCLEAN -SN immediately before starting dccd.

`dbclean -SN` would have emptied the database.
As checksums flooded in, the hash table for the database would have filled.  
Normally, a timer checks to see if the hash table is nearly full and starts
`dbclean` to rebuild database with a bigger hash table.  Evidently in
this case by the timer fired, the hash table was too full and there was not
enough room to add the checksum for the server-ID claim.  As a result,
`dbclean` was run in an emergency repair mode.

Vernon Schryver

More information about the DCC mailing list

Contact by mail or use the form.