OnLine 4.10 ABORTs - Reason?

Recently one of our OnLine instances aborted, leaving the following information in the turbo.log file: =============================== I/O bad request chunk 0, pagenum 0, pagecnt 1 partition header error: 49cdfc: 00000000 00000000 00000000 00830000 ........ ........ [hex dump followed] -- Fail Consistency Check -- pthdrpage:ptwrite:bad bfget -- pid=19318 user=200 us=301c3c Core for 19318 in '/home/utilities/data/Spooler' Process Aborted Abnormally: pid=19318 user=200 us=301c3c flags=1 Process Aborted Abnormally (latch): pid=19318 use=200 flags=1 =============================== This would imply that the user structure at 301c3c in shared memory was in use, and that the process (19318) was holding a latch. As far as I can tell, the process was not terminated forcefully (kill -9). I'm not sure what chunk 0 is, as the chunks are numbered from 1 (holds rootdbs) upwards. Even stranger, exactly the same thing happened half an hour after the database had been brought back on line. The user structure address was the same. The database hasn't misbehaved since then, so I wonder what caused two similar errors in such a short space of time... BTW the machine is a Sun Sparc 20/51 running Solaris SunOS 4.1.3_U1. It would be interesting to hear any suggestions... -- Richard Williams Email: rdw@reflex.co.uk Reflex Manufacturing Systems Tel: (0293) 560111