[zombor]

I rebooted the server tonight, and all my free space came back, so du was telling me the right amount. I will have to wait and see if the problem persists.

[zombor]

[QUOTE][u]Quote by: khiltd[/u][p]It’s difficult to see what it is we’re supposed to be comparing there since I’m not apt to grab a calculator and work out the percentages myself, but if you’re seeing a discrepancy it could be a difference in the way block sizes are being calculated.
[/p][/QUOTE]

df -h (and the finder) says my usage is 75G. du (and another tool called WhatSize) tell me my usage is 33G. I’m trying to find where all the space is being used up, as it keeps going up. Last Friday I had about 10G free, now this morning i have less than 1G. My thoughts were that a log file was being run up somewhere, but du doesn’t see m to tell me where the space is being used up, because it tell me there is only 33G used on the disk. I’ve never seen df and du tell me different results.

[zombor]

Seems the problem was with flock()…I did a ktrace on the httpd process and there were a ton of flock() calls taking half a second each. I swithced G2 from file locking to database locking and the load time cut in half, however still slower than it should be.

I did some more ktraces, and now i get a bunch of lseek()’s taking .1-.3 seconds each, kdump output is:

7962 httpd 0.125575 CALL lseek(0xa,0,0)
7962 httpd 0.126855 CALL close(0xa)
7962 httpd 0.127335 RET close 0
7962 httpd 0.128935 RET getdirentries 664/0x298
7962 httpd 0.132929 CALL lseek(0xa,0,0)
7962 httpd 0.135916 CALL lseek(0xa,0,0)
7962 httpd 0.138924 RET flock 0
7962 httpd 0.155503 CALL access(0x2875400,0)
7962 httpd 0.160541 CALL lseek(0xa,0,0)
7962 httpd 0.165998 CALL lseek(0xa,0,0)
7962 httpd 0.182390 CALL lseek(0xa,0,0)
7962 httpd 0.186547 CALL lseek(0xa,0,0)
7962 httpd 0.443527 GIO fd 9 read 4 bytes

Anyone have any other info I could try out?

[Author]

Posts