On 09/20/2011 05:28 AM, Igor Galić wrote:
Right now the last thing left is iocore/ And a second pass to see if I didn't leave anything out. I have halted the progress however, because Leif reported segfaults with no traces with current trunk as forward proxy. I've been running the server compiled with --enable-debug for several hours and can't confirm - can somebody else? i
What seems to happen is that if I restart the server, once in a while (still haven't figured out how to reliably reproduce it), the cache gets corrupted somehow, and I get an error e.g.
Sep 20 09:00:19.924] Server {0x2ae1fb40f700} WARNING: cache disk operation failed READ -1 14 [Sep 20 09:00:19.924] Server {0x2ae1f8f17700} WARNING: disk read error on recover '/dev/disk/by-id/scsi-SATA_Maxtor_6B300S0_B612SBPH 65536:36623970', clearing
When this happens, there's also a chance it will segfault. I've never seen this before, but now it seems to have fairly frequently. Which kinda makes sense, assuming 14 is EFAULT. I'll investigate some more.
-- Leif
