Re: Postmaster hangs

On 31/10/2009 4:32 PM, Karen Pease wrote:
> Sorry for the delay in responding, and thanks for your help.

By the way, you may also be able to learn some more using the 'blktrace'
tool. This gathers low-level data about what processes are performing
I/O on your system.

# mount -t debugfs none /sys/kernel/debug
# btrace /dev/YOUR_DISK

eg on my workstation, Pg is busy on /dev/md1 so:

# btrace /dev/md1
> 9,1 0 13496 2.514811417 31109 A W 114959520 + 8 <- (252,3) 31073056
> 9,1 0 13497 2.514811732 31109 Q W 114959520 + 8 [postgres]
> 9,1 0 13498 2.514816413 31109 A W 114959528 + 8 <- (252,3) 31073064
> 9,1 0 13499 2.514816732 31109 Q W 114959528 + 8 [postgres]
> 9,1 0 13500 2.514819960 31109 A W 114959536 + 8 <- (252,3) 31073072
> 9,1 0 13501 2.514820280 31109 Q W 114959536 + 8 [postgres]
> 9,1 0 13502 2.514823399 31109 A W 114959544 + 8 <- (252,3) 31073080
> 9,1 0 13503 2.514823737 31109 Q W 114959544 + 8 [postgres]
> 9,1 0 13504 2.514826792 31109 A W 114959552 + 8 <- (252,3) 31073088
> 9,1 0 13505 2.514827112 31109 Q W 114959552 + 8 [postgres]
> 9,1 0 13506 2.514830715 31109 A W 114959560 + 8 <- (252,3) 31073096
> 9,1 0 13507 2.514831034 31109 Q W 114959560 + 8 [postgres]
> 9,1 0 13508 2.514834278 31109 A W 114959568 + 8 <- (252,3) 31073104
> 9,1 0 13509 2.514834593 31109 Q W 114959568 + 8 [postgres]
> 9,1 0 13510 2.514838977 31109 U N [postgres] 0
> 9,1 0 13511 2.515773952 31109 U N [postgres] 0
> 9,1 0 13512 2.515778889 16 U N [kblockd/0] 0
> 9,1 0 13513 2.519146885 31109 U N [postgres] 0
> 9,1 0 13514 2.520511306 31109 U N [postgres] 0
> CPU0 (9,1):
> Reads Queued: 1, 4KiB Writes Queued: 6718, 26872KiB
> Read Dispatches: 0, 0KiB Write Dispatches: 0, 0KiB
> Reads Requeued: 0 Writes Requeued: 0
> Reads Completed: 0, 0KiB Writes Completed: 0, 0KiB
> Read Merges: 0, 0KiB Write Merges: 0, 0KiB
> Read depth: 0 Write depth: 0
> IO unplugs: 53 Timer unplugs: 6

(ignore quotes; working around a stuipid Thunderbird limitation)

--
Craig Ringer