Hi,

> Well yes.  I suspect the inode hashtable could be shrunk; I don't think we
> do hash-based inode lookups very much?

Id be interested to know how often we hit the inode hash. Last time I
looked at it, there were some real deep chains due (Im guessing) to
the way ext2 inode allocation causes them to end up in groups.

The attached graph shows what is going on. The histogram below shows its
not as bad as I thought, most of the distribution is in the 0-1 bucket
range.

> It would be very nice to have some confirmation that the size of these
> tables is being appropriately chosen, too.  Maybe we should shrink 'em 32x
> and see who complains...

Why dont we just do node round robin allocations during boot? This
should mean the static boot time hashes would at least end up on
different nodes.

Anton

0 248652
1 7374
2 77
3 71
4 72
5 109
6 61
7 75
8 114
9 54
10 34
11 45
12 58
13 35
14 37
15 39
16 32
17 30
18 36
19 31
20 31
21 43
22 47
23 52
24 30
25 26
26 31
27 35
28 35
29 38
30 39
31 38
32 43
33 40
34 49
35 43
36 41
37 44
38 47
39 49
40 51
41 67
42 66
43 104
44 166
45 195
46 289
47 372
48 467
49 542
50 589
51 516
52 436
53 271
54 100
55 53
56 22
57 0
58 1