On Mon, Jun 27, 2016 at 08:09:58AM +0200, Sven Eckelmann wrote:
On Monday 27 June 2016 06:02:18 Linus Lüssing wrote:
[...]
Looks good. Do you have any typical numbers
regarding tt_change_cache
and tt_req_cache sizes for specific setups?
No, I don't have any numbers.
Freifunk Rhein-Neckar at about 4pm (~700 nodes, ~1800 clients, batman-adv compat15):
-----
batadv_tt_roam_cache 0 0 64 63 1 : tunables 120 60 8 : slabdata
0 0 0
batadv_tt_req_cache 1 63 64 63 1 : tunables 120 60 8 : slabdata
1 1 0
batadv_tt_change_cache 0 0 32 124 1 : tunables 120 60 8 : slabdata
0 0 0
batadv_tt_orig_cache 3210 3213 64 63 1 : tunables 120 60 8 : slabdata
51 51 0
batadv_tg_cache 3162 3162 128 31 1 : tunables 120 60 8 : slabdata
102 102 0
batadv_tl_cache 2 31 128 31 1 : tunables 120 60 8 : slabdata
1 1 0
-----
-> snapshot taken from an x86 machine connected to a Gluon ar71xx node.
So the roam, req, change and tl cache are mostly empty here. On the other hand, they
could
probably gain a significant size once a larger LAN gets bridged to bat0, I guess.
Also looks like various file systems have about half a dozen caches each with most
of them being quite empty most of the time.
So, I guess it's ok to have the roam/req/change/tl caches.
Not sure whether I should (since TT is Antonio's part) but the patch looks
correct, the previous version with the same concept tested and looks very
similar to what I had incidentally been coding at the same time as Sven :), so:
Acked-by: Linus Lüssing <linus.luessing(a)c0d3.blue>