Description of Problem:
Running cerberus 9-20 against an Dell PERC3/QC (AMI MegaRAID) controller,
there are a bunch of SCSI read/write timeouts. AMI recommends increasing
the SD_TIMEOUT value from 30 seconds to 60 seconds.
occasional. Always with enough disk activity. Seen on kernel 2.4.7-0.3.
Steps to Reproduce:
1. Run cerberus 9-20 against a PERC3/QC
see SCSI disk timeouts in dmesg
no disk timeouts reported
Patch to increase timeout.
--- linux-2.4/drivers/scsi/sd.c.orig Fri Aug 3 17:27:57 2001
+++ linux-2.4/drivers/scsi/sd.c Fri Aug 3 17:28:05 2001
@@ -81,7 +81,7 @@
* Time out in seconds for disks and Magneto-opticals (which are slower).
-#define SD_TIMEOUT (30 * HZ)
+#define SD_TIMEOUT (60 * HZ)
#define SD_MOD_TIMEOUT (75 * HZ)
struct hd_struct *sd;
If the kernel team feels this is SHOULD- or MUST-FIX, we can mark it as such.
SHOULD FIX and done so.