> If after it decreases in size it stays there it might be similar to:
>         7111576 arc shrinks in the absence of memory pressure

After it dropped, it did build back up. Today is the first day that
these servers are working under real production load and it is looking
much better. arcstat is showing some nice numbers for arc, but l2 is
still building.

read  hits  miss  hit%  l2read  l2hits  l2miss  l2hit%  arcsz  l2size
 19K   17K  2.5K    87    2.5K     490    2.0K      19   148G    371G
 41K   39K  2.3K    94    2.3K     184    2.1K       7   148G    371G
 34K   34K   694    98     694      17     677       2   148G    371G
 16K   15K  1.0K    93    1.0K      16    1.0K       1   148G    371G
 39K   36K  2.3K    94    2.3K      20    2.3K       0   148G    371G
 23K   22K   746    96     746      76     670      10   148G    371G
 49K   47K  1.7K    96    1.7K     249    1.5K      14   148G    371G
 23K   21K  1.4K    93    1.4K      38    1.4K       2   148G    371G

My only guess is that the large zfs send / recv streams were affecting
the cache when they started and finished.

Thanks for the responses and help.

zfs-discuss mailing list

Reply via email to