Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
Hi Erik, sorry, I didn't mention that you have to search for your process name (the one in braces at the end). The start point is always the queuing "Q", then it is possible that another process does the dispatch "D" (sending to the driver) for you and again another process handles the completions "C" afterwards. If you use "dd" with "oflag=direct", then the "dd" process handles that all by itself. A full cycle for a single IO can look like this: 8,0 0 41 5.467391803 14390 A WS 94672896 + 1024 <- (8,2) 94281728 8,0 0 42 5.467400603 14390 Q WS 94672896 + 1024 [dd] 8,0 0 43 5.467404515 14390 G WS 94672896 + 1024 [dd] 8,0 0 44 5.467410381 14390 P N [dd] 8,0 0 45 5.467413803 14390 I WS 94672896 + 1024 [dd] 8,0 0 46 5.467415759 14390 U N [dd] 1 8,0 0 47 5.467418203 14390 D WS 94672896 + 1024 [dd] 8,0 0 48 5.469839670 0 C WS 94672896 + 1024 [0] With the sector number (here "94672896") you can see that this is all the same IO. Other answers inline. Cheers, Sebastian On 27.09.2012 11:37, Erik Schwalbe wrote: > dd if=/dev/zero of=/dev/data/test bs=1M count... > 251,12 4 10921 0.107595919 1805 Q W 275888 + 8 [flush-251:12] This must be a file system flush. This is not your "dd" process > dd if=/dev/zero of=/dev/data/drbd bs=1M count... > 251,2 3 269678 12.232806819 12635 Q WS 2078359544 + 8 [nfsd] This is your NFS daemon - also the wrong process. > Please can you explain, what that mean?? > And why there is only +8. Especially FS operations do from time to time small 4 KiB flushes. This is normal behaviour.