OpenVZ Forum


Home » General » Support » Strange memory usage: same VPS but two different HNs
Strange memory usage: same VPS but two different HNs [message #28483] Thu, 20 March 2008 09:23
draga is currently offline  draga
Messages: 20
Registered: July 2007
Junior Member
Hello everybody. Follows a ticket I've opened on my new external OpenVZ VPS provider's support system. I'd like also to put it here because I think it may be interesting and because I think someone could have some ideas on what's happening:

Good morning.
This morning I've found some services down due to a privvmpages fail.
So, I've decided to copy all the extenal VPS data to the original HN in which the dragas VPS has been running until yesterday, disconnect it from the net and keep an eye on both of them.
After some hours, I can see that the total RAM occupation of my HN (called dragasepia, a Via Epia with 512MB Ram) running a copy of my external VPS is:
dragasepia:~# free
total used free shared buffers cached
Mem: 498972 464728 34244 0 15312 163192
-/+ buffers/cache: 286224 212748
Swap: 1171792 60 1171732

So, my VPS is using more or less 250MB of Ram, considering 30 MB for the HN.
The /proc/userbean_beancounters of my VPS copy inside my HN is:
dragas:/# cat /proc/user_beancounters
Version: 2.5
uid resource held maxheld barrier limit failcnt
2000: kmemsize 6030356 8051778 40098656 44108521 0
lockedpages 0 0 32 32 0
privvmpages 178367 225531 2147483647 2147483647 0
shmpages 0 336 8192 8192 0
dummy 0 0 0 0 0
numproc 122 164 10000 10000 0
physpages 77492 79751 0 2147483647 0
vmguarpages 0 0 2147483647 2147483647 0
oomguarpages 77492 79751 6144 2147483647 0
numtcpsock 31 56 2000 2000 0
numflock 19 23 100 110 0
numpty 1 2 16 16 0
numsiginfo 0 3 256 256 0
tcpsndbuf 276272 519856 9000000 9000000 0
tcprcvbuf 507904 655720 5120000 5120000 0
othersockbuf 187152 249356 5120000 5120000 0
dgramrcvbuf 0 35040 262000 262000 0
numothersock 140 174 2000 2000 0
dcachesize 0 0 1966080 1966080 0
numfile 2917 3666 15000 15000 0
dummy 0 0 0 0 0
dummy 0 0 0 0 0
dummy 0 0 0 0 0
numiptent 80 80 400 400 0

Now let's get back to the external VPS.
If I ask a "free" (but, this time, I have to ask it from the VPS since I cannot access the HN), I get:
dragas:/tmp# free
total used free shared buffers cached
Mem: 1572864 757912 814952 0 0 0
-/+ buffers/cache: 757912 814952
Swap: 0 0 0

and the /proc/userbean_beancounters is:
dragas:/tmp# cat /proc/user_beancounters
Version: 2.5
uid resource held maxheld barrier limit failcnt
984470: kmemsize 11321260 14893534 2147483646 2147483646 0
lockedpages 0 5 999999 999999 0
privvmpages 186662 393224 393216 393216 5
shmpages 640 2576 196608 196608 0
dummy 0 0 0 0 0
numproc 110 175 999999 999999 0
physpages 108107 127126 0 2147483647 0
vmguarpages 0 0 196608 2147483647 0
oomguarpages 108107 127126 196608 2147483647 0
numtcpsock 31 79 7999992 7999992 0
numflock 34 40 999999 999999 0
numpty 1 4 500000 500000 0
numsiginfo 0 14 999999 999999 0
tcpsndbuf 635376 1281968 161060864 343087104 0
tcprcvbuf 507904 1337840 161060864 343087104 0
othersockbuf 245376 1535680 161060864 343087104 0
dgramrcvbuf 0 103552 161060864 343087104 0
numothersock 183 296 7999992 7999992 0
dcachesize 835045 1039473 2147483646 2147483646 0
numfile 3099 4905 23999976 23999976 0
dummy 0 0 0 0 0
dummy 0 0 0 0 0
dummy 0 0 0 0 0
numiptent 14 71 999999 999999 0


as you can see, the privvmpages counts 5 fails.
Moreover, please note that the VPS inside my HN is also running Apache Tomcat while I had to turn it off in my external VPS because it raised the Ram occupation, causing a fail in privvmpages.

There must be something wrong, it seems quite strange that the same VPS is using 250 MB of ram in a HN and 800 MB in another. More than 900 MB if starting Tomcat.
Previous Topic: Vzmigrate bug?
Next Topic: Apache2 Memory usage so high(280MB)
Goto Forum:
  


Current Time: Mon Dec 05 01:36:39 GMT 2022

Total time taken to generate the page: 0.02193 seconds