Alpha V8.3 SAMBA CIFS V1.1-1 ECO1 - not enough core error messages

[Peter Barkas]

SAMBA CIFS is configured as a simple standalone server as documented in a previous thread.

The SMBD log contains these messages:

...

[2010/01/30 00:43:23, 0] SAMBA$SRC:[SOURCE.LIB]DEBUG.C;1:(625)
Unable to open new log file /samba$root/var/log_kronos.smbd: not enough core

...

[2010/01/30 00:48:04, 0] SAMBA$SRC:[SOURCE.SMBD]SERVICE.C;1:(1009)
'SAMBA$ROOT:[TMP]' does not exist or permission denied when connecting to [IPC$] Error was not enough core

...

These messages caused errors on the Windows Server client.

The SAMBA CIFS server had been working successfully without logging these errors since it was started about 3 days before the errors were logged.

The cause and therefore solution is not clear to me.

[Peter Barkas]

Probably, the error occurred was because the account in which SAMBA was running did not have delete privilege for the log file.

The server had been restarted following an earlier problem and (inadvertently) was not running in the SYSTEM account.

It appears that SAMBA may have determined that it was time for a new log file even though the space used in the existing file was well below the max logfile threshold.

It seems that the threshold may be checked against the allocated rather than used space and the SAMBA service startup allocates a very generous RMS extendsize.

[Peter Barkas]

10 points for finding the solution, -20 points for causing it.