128 Buckets Failures

[romeor]

Hello dear All,

i searched the forum and didnt find anything interesting on my question.
the question is

[root@gw2 ~]# vmstat -z
ITEM                     SIZE     LIMIT      USED      FREE  REQUESTS  FAILURES
64 Bucket:                536,        0,      193,        3,      193,      159
128 Bucket:              1048,        0,     1956,        0,     1956,    10418

So this is seems like mem allocation failures for smth. how i can cure it? i've got 2 routers with freebsd 8.2 and PF running on them. and both have those failures.

Mem: 72M Active, 942M Inact, 429M Wired, 320K Cache, 408M Buf, 2404M Free
Swap: 5900M Total, 5900M Free
CPU:  0.0% user,  0.0% nice,  0.1% system,  0.1% interrupt, 99.8% idle

what should i do?
amd64 on both.

 

[Terri_Kennedy]

the question is

[root@gw2 ~]# vmstat -z
ITEM                     SIZE     LIMIT      USED      FREE  REQUESTS  FAILURES
64 Bucket:                536,        0,      193,        3,      193,      159
128 Bucket:              1048,        0,     1956,        0,     1956,    10418

So this is seems like mem allocation failures for smth. how i can cure it? i've got 2 routers with freebsd 8.2 and PF running on them. and both have those failures.

Are you experiencing some operational issue which led you to suspect the zone allocator?

I'm seeing improbably high numbers here as well:

(0:17) gate:/tmp# uname -pr
6.4-STABLE i386
(0:18) gate:/tmp# vmstat -z
ITEM                     SIZE     LIMIT      USED      FREE  REQUESTS  FAILURES
64 Bucket:                268,        0,       19,       37,      107,       11
128 Bucket:               524,        0,     2841,        1,     7862, 21391388

(0:67) new-gate:/tmp# uname -pr
8.2-STABLE amd64
(0:68) new-gate:/tmp# vmstat -z
ITEM                     SIZE     LIMIT      USED      FREE  REQUESTS  FAILURES
64 Bucket:                536,        0,      154,        0,      154,        1
128 Bucket:              1048,        0,     1723,        2,     1723,   434426

(Only lines with non-zero failure counters shown.)

I din't see how it is possible to have 2700+ and 250+ times as many failures as requests, unless failures are not related to requests.

Unless this is causing some actual impact on your system, I wouldn't worry about it - perhaps one of the implementors will drop by and explain what the failure counter really means in this report.

 

[romeor]

The "REQUESTS" tab actually shows successful requests, not the overall requests number.
i just like when my system runs cleanly, without some unknown window$-like errors, so it is still interesting to me, whats wrong and if it is expected to be so, than what does it mean and why such situation may come up

Are you experiencing some operational issue which led you to suspect the zone allocator?

actually not. today i just seen that traffic between vlans aint going higher, that 50 mbits (while card and switch port are 1G) so it was interesting to me if these two are related to each other. while testing the wan speed i've got expected traffic.