Forum Discussion

imac_105647's avatar
imac_105647
Icon for Nimbostratus rankNimbostratus
Mar 31, 2010

BigIP swapping

Hello,

 

 

I asked this on the random question forum, but this is perhaps a better place:

 

 

We have a pair of LTMs 3600 running in Active/Standby HA mode and 9.4.8

 

 

We have had issues when applying our ASM policy where the HA fails over and back again which is described here: SOL10095

 

 

Looking at the machines from the command line I see what appears to show that the machines are swapping. This would seem like a Bad Thing, but I am told that BigIPs will swap, what I don't understand is:

 

 

a) Is top lying to me and they aren't swapping?

 

 

b) Should the boxes be swapping?

 

 

c) Why would the standby box swap immediately after a reboot as it is doing very little?

 

 

Here is top from the Active:

 

 

12:54:16 up 202 days, 18:42, 2 users, load average: 1.08, 1.12, 1.09

 

220 processes: 218 sleeping, 2 running, 0 zombie, 0 stopped

 

CPU states: cpu user nice system irq softirq iowait idle

 

total 54.1% 0.8% 0.8% 0.0% 0.0% 0.0% 44.1%

 

cpu00 8.3% 1.6% 1.6% 0.0% 0.0% 0.0% 88.3%

 

cpu01 100.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%

 

Mem: 3332068k av, 3285924k used, 46144k free, 0k shrd, 104720k buff

 

422332k actv, 41300k in_d, 50148k in_c

 

Swap: 6352892k av, 1010856k used, 5342036k free 149896k cached

 

 

Here is top from the Standby:

 

 

12:55:14 up 1 day, 2:12, 1 user, load average: 1.13, 1.07, 1.02

 

208 processes: 206 sleeping, 2 running, 0 zombie, 0 stopped

 

CPU states: cpu user nice system irq softirq iowait idle

 

total 50.7% 0.0% 0.0% 0.0% 0.0% 0.0% 49.2%

 

cpu00 1.9% 0.0% 0.0% 0.0% 0.0% 0.0% 98.0%

 

cpu01 100.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%

 

Mem: 3332068k av, 3319456k used, 12612k free, 0k shrd, 52760k buff

 

427796k actv, 51708k in_d, 62156k in_c

 

Swap: 6352892k av, 498708k used, 5854184k free 255880k cached

 

 

Here is me confused:

 

 

:crazy:
application delivery
design
performance
testing

4 Replies

Sort By
  • hooleylist's avatar
    hooleylist
    Icon for Cirrostratus rankCirrostratus
    Apr 01, 2010
    Hi Ian,

     

     

    I'm seeing something similar on a lab 3600 with ASM licensed and no load:

     

     

    13:46:57 up 23:22, 1 user, load average: 1.14, 1.03, 1.01

     

    216 processes: 213 sleeping, 3 running, 0 zombie, 0 stopped

     

    CPU states: cpu user nice system irq softirq iowait idle

     

    total 51.8% 0.3% 0.0% 0.0% 0.0% 0.0% 47.5%

     

    cpu00 3.6% 0.8% 0.2% 0.0% 0.0% 0.2% 95.2%

     

    cpu01 100.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%

     

    Mem: 3332068k av, 3283896k used, 48172k free, 0k shrd, 60804k buff

     

    428020k actv, 47752k in_d, 44728k in_c

     

    Swap: 6352892k av, 515388k used, 5837504k free 201080k cached

     

     

    I'll check with Chris on this and see if we can get more info on why setting the ASM policy active would affect the unit so badly.

     

     

    Aaron
  • imac_105647's avatar
    imac_105647
    Icon for Nimbostratus rankNimbostratus
    Apr 06, 2010
    Thanks Aaron,

     

     

    I'm surprised others aren't seeing similar things with their 3600s.

     

     

    Ian
  • Hamish's avatar
    Hamish
    Icon for Cirrocumulus rankCirrocumulus
    May 11, 2010
    Ignore top, and run vmstat... e.g.

     

     

    vmstat 10

     

     

    and ignoring the first line that comes out, e.g.

     

     

    vmstat[root@vltm-mb:Active] config vmstat 5

     

    procs -----------memory---------- ---swap-- -----io---- --system-- -----cpu------

     

    r b swpd free buff cache si so bi bo in cs us sy id wa st

     

    0 0 104 105620 83296 212228 0 0 1 2 39 78 1 0 98 0 0

     

    1 0 104 105568 83308 212244 0 0 0 19 358 1940 1 0 96 3 0

     

     

    will show the vmstats (Virtual Memory Stats) every 5 seconds. The first line is the stats since boot. The rest are offsets from the previous line.

     

     

    Unless the system is actively swapping, there's not a lot to worry about, UNLESS your sod daemon starts to get paged. If sod pages, you'll get failovers, active/active (split brain) etc... All sorts of hassels. (In short if it's a standalone system, don't worry till it pages, if it's an HA pair, you need to take action, possibly by increasing the amount of memory allocated to the linux partition. I'm not sure whether you still have to edit the grub config file after mounting the /boot filesystem with 9.4.8 or not. I suspect you do).

     

     

    H
  • Cspillane_18296's avatar
    Cspillane_18296
    Icon for Nimbostratus rankNimbostratus
    May 26, 2010
    Hi all,

     

     

    Just for info, the swapping (and other issues such as both units in the HA pair becoming active under certain circumstances) was resolved by provisioning more memory:

     

     

    b db provision.extramb

     

     

     

    Chris