Load average - 4.07, 1.73, 0.69, why the load so HIGH at just 20-30mbps?

genic

2.1-RC0 (i386)
built on Fri Jun 28 15:03:42 EDT 2013
FreeBSD 8.3-RELEASE-p8

Platform nanobsd (4g)
NanoBSD Boot Slice pfsense1 / ad0s2 (ro)
CPU Type Intel(R) Pentium(R) III CPU family 1400MHz
Hardware crypto SafeNet SafeXcel-1141 rng des/3des aes md5 sha1 null

State table size 1017/51000
MBUF Usage 291/16704
Load average 4.07, 1.73, 0.69

this wasnt a problem until recent nightly updates. i do not have packages installed other than dhcpd, dnsmasq, ntpd. just a couple rules…and all the traffic is both tcp/udp...i am updating to the latest nightly now to see if it goes away, but for some reason the cpu just spikes to 100% constant until i reboot.

anyone have any ideas? previous load what very low even when traffic was higher (50mbps+)

genic

updated to the latest nightly and the cpu spike seems to be fixed. strange. it was in the june 28th nightly that was causing the problems i think.

Load average – 0.17, 0.13, 0.13 ....pretty constant. @ at around ~30mbps

tks to whatever fixed.

jimp

For future reference, load alone is pretty meaningless. To even speculate about a cause, we'd need to see what processes are actually running from something like "top -SH".

genic

@jimp:

For future reference, load alone is pretty meaningless. To even speculate about a cause, we'd need to see what processes are actually running from something like "top -SH".

gotcha bossman. thanks for the help, and thanks for the work on all the work for this project.

i will provide that info in the future, but i am not as *ix smart as 99.999% as the rest of people who are on this forum.

:-[

genic

and other activities (more on those later…i will amend this post with those, but not at the moment.)

please see the attached photo for the system activities after i backup (the backup finishes fine..then tries to reload and gets 'stuck' and i have reboot)

any ideas?

the nightly is as follows:
2.1-RC0 (i386)
built on Thu Jul 4 03:04:50 EDT 2013
FreeBSD 8.3-RELEASE-p8

nanobsd - (4g)
NanoBSD Boot Slice - pfsense1 / ad0s2 (ro)
CPU Type - Intel(R) Pentium(R) III CPU family 1400MHz
Hardware crypto - SafeNet SafeXcel-1141 rng des/3des aes md5 sha1 null (not enabled within pfsense)

these hardware boxes are the: watchguard firebox 700 (i have several)

also, as you can see i dont have many packages running on here 'yet' ... i am just trying to get them solid before i configure them for failover.

sys_activity_after_backup.jpg_thumb

wallabybob

I think there is something not right in check_reload_status using over 44% and dhclient and dhcpd each using over 10%.

Is your WAN interface (or some other interface getting its IP address by DHCP) repeatedly going up and down?

Have you looked in the system log for "strange" behaviour? (See Status -> System Logs)

genic

the WAN isnt going down, and there is nothing strange in the system logs and/or other logs. it just seems the check_reload_status goes bonkers and then shuts down the dns forwarder service and dhcp service after a while. the only fix is to reboot.

i found that is not just the bandwidth that causes this. it happens when i do different system functions like a backup along with other system functions that cause the check_reload_status.
its strange because it started to happen on the nightlies around the end of june (i update to a new nightly each day in this environment since its my test lab) and has affected each nightly in the month of june.

…so i dont know what other information i can output for everyone to help diagnose this problem. if you could direct me to what you might want/need to help...i would be more than welcome to gather that information.

however…on the other side of the coin.

….i use the RC build on the same exact machine configurations in twelve other colo's that range from a tower boxes to pretty large caged locations/office data centers that have many different networks and many racks stacked full of diverse equipment. those machines run solid, but even with load balancing i am running up against the 100mpbs port limitations on the boxes (and VPN power) .. (if they had 1gbps ports..i would be so very happy, but then i would run out of cpu/memory/storage on the boxes before i could utilize the full 1gbps.)
so i am going to move away from the watchguards to spec'd 2u machines from dell that i am currently working on rolling out.
sorry, i know none of that means anything to the original post, but i like to promote the use of these boxes (along with pfsense) to SMB's because they are just the little boxes that can!