scp stalls after 800 MB while disk util goes to 100%

[Fredrik Uddin-Backman2]

Hi

I have a new BL860i2 wiuth latest 11.31 1009 installed. When copying a file from a second node over dedicated 1GBe my transfer stalls after some hundred MB. At same time I see my disk util in Glance goes to 100% for a few seconds. After that scp resumes and finishes normally. This happends to internal disks as well as to XP24000 luns. Funny thing is if I copy same file again this does not happen. Copy same file but to new filename, same thing happens, stalling. Even scp to localhost gets stalled Any suggestions?

[vishnu.khandare]

Hi Fredrik,

Check what are other processes running at that time while ur doing scp, check disk util by sar command also.


Regards
Vishnu Khandare

[Chandrahasa s]

Hi,

run sar -d 2 5

run iostat 2 2

while doing scp

Chandra

[Fredrik Uddin-Backman2]

fcache_fb_policy=1 or fcache_fb_policy=2 completly changes the picture here. It seems like default fcache_fb_policy=0 give the system a strange behaviour.

sar output with fcache_fb_policy=0 attached. It seems like when system flushes cache, it stops receiving to cache for a while...