#4241 Wrong real memory after update

[Anonymous]

Hi, I recently upgraded Webmin to 1.630
It appears that the RAM calculation has broken completely.
I run CentOS 6.4 on a OpenVZ kernel.¨

Output from Webmin panel:
"Real memory: 256 MB total, 173.89 MB used"

Output from free -m:
" total used free shared buffers cached
Mem: 1024 704 319 0 0 0 "

[Anonymous]

Picture...

[Jamie Cameron]

Could you post or attach the /proc/user_beancounters file on your system? Webmin 1.630 uses the privvmpages line from that file to work out the memory limit under OpenVZ.

[Jon Bogan]

Not the same user as above but seeing the same issue.

Currently on Webmin 1.650 and I guess I didn't notice this after the 1.630 upgrade.

Kernel and CPU Linux 2.6.32-042stab079.5 on x86_64
Processor information Intel(R) Xeon(R) CPU X3220 @ 2.40GHz, 4 cores

Attached similar screenshot plus my beancounters:

[root@bandit ~]# cat /proc/user_beancounters | grep privvmpage
privvmpages                   261               173192  9223372036854775807  9223372036854775807                    0
privvmpages                 51005               230880               262144               262144                    3
privvmpages                 45987               204336               262144               262144                    3
privvmpages                103452               750000               500000               750000                  246
privvmpages                372420               439518  9223372036854775807  9223372036854775807                    0

This is on the host node. The containers seems to report memory correctly.

[Jamie Cameron]

Jon - could you also post the contents of your /proc/meminfo file?

[Jon Bogan]

Here you go, Jamie:

[root@bandit ~]# cat /proc/meminfo
MemTotal:        8041728 kB
MemFree:         5493076 kB
Buffers:          317896 kB
Cached:          1087244 kB
SwapCached:            0 kB
MemCommitted:    4456448 kB
VirtualSwap:           0 kB
Active:          1482760 kB
Inactive:         815660 kB
Active(anon):     699000 kB
Inactive(anon):   194748 kB
Active(file):     783760 kB
Inactive(file):   620912 kB
Unevictable:           0 kB
Mlocked:               0 kB
SwapTotal:       6160376 kB
SwapFree:        6160376 kB
Dirty:              8004 kB
Writeback:             0 kB
AnonPages:        893440 kB
Mapped:            66612 kB
Shmem:               484 kB
Slab:             170080 kB
SReclaimable:     134916 kB
SUnreclaim:        35164 kB
KernelStack:        3536 kB
PageTables:        22236 kB
NFS_Unstable:          0 kB
Bounce:                0 kB
WritebackTmp:          0 kB
CommitLimit:    10181240 kB
Committed_AS:    1948184 kB
VmallocTotal:   34359738367 kB
VmallocUsed:      108224 kB
VmallocChunk:   34359625808 kB
HardwareCorrupted:     0 kB
AnonHugePages:         0 kB
HugePages_Total:       0
HugePages_Free:        0
HugePages_Rsvd:        0
HugePages_Surp:        0
Hugepagesize:       2048 kB
DirectMap4k:        9976 kB
DirectMap2M:     8378368 kB

[Jamie Cameron]

Ok, I see the problem - for some reason, your /proc/user_beancounters file has multiple privvmpages lines! Webmin doesn't expect this, and is using the last one to get the RAM limit, which is clearly wrong. I will fix this in the next Webmin release.

The code change that fixes it can be seen at : https://github.com/webmin/webmin/commit/6ed31c6d1c453d42660182a933b2d3ba7f29eefc

[Jon Bogan]

That's because this is the host node. Individual containers have only one but the host node contains the bean counters for all containers. The host node is "0" and is last in my /proc/user_beancounters but I don't know for sure if that's always the case.

Thanks for the quick turnaround!

Jon