Memory not updated

[RenatoMN]

Hello. I created a VPS. Then I needed to increase the allocated memory and swap (from 2gb to 3gb). I edited the fields, clicked save, and nothing happens on the VM.

Proxmox screen now show on bottom:

Resource          / Current / Maximum
Memory/Swap (MB)     / 1565       / 3072

But inside VM, running top command, i still see only 2GB memory total...

tried reboot, and still the same. tried also stop it, change settings, start again, and the problem persists...

i have lxadmin on the vm, may it be something related to lxadmin?

thanks

 

[tom]

Hello. I created a VPS. Then I needed to increase the allocated memory and swap (from 2gb to 3gb). I edited the fields, clicked save, and nothing happens on the VM.

Proxmox screen now show on bottom:

Resource          / Current / Maximum
Memory/Swap (MB)     / 1565       / 3072

But inside VM, running top command, i still see only 2GB memory total...

tried reboot, and still the same. tried also stop it, change settings, start again, and the problem persists...

i have lxadmin on the vm, may it be something related to lxadmin?

thanks

I just tried that with a Debian template, "top" shows immediately the increased value inside the container. what template do you use?

 

[RenatoMN]

i use the centos5 provided on proxmox appliances download list

-edit-

the steps i followed:

create the vm
add 1 more ip
run vzcfgvalidate -i /etc/vz/conf/101.ini
it suggested me to set the vm barrier to 0 (i agreed) -- an admin (cant remember if it was you) said proxmox team will fix it on next update, seems its not.
started it
yum update
yum upgrade
then downloaded lxadmin
started installation

i did nothing using terminal after it, outside or inside vm, i only opened lxadmin panel, added a domain and uploaded files into it (and restored the database)
the memory use raised when I unlocked the website forum, then i needed to increase it. then the above problem occured.

sorry my english

-edit 2-

found the thread about the physpages barrier

http://www.proxmox.com/forum/showthread.php?t=777&highlight=barrier

 

[RenatoMN]

sorry to double post, but i'm facing some strange issues...

proxmox home screen show 2gb/400mb (400mb of 2gb in use)
vm screen show 85mb use

'top' on proxmox show me almost all memory in use (50mb free)
'top' on vm show 85mb use

but lighttpd sometimes give us "Internal Server Error", as a insuficient memory error we had on last host. we installed it over proxmox because we will need 2 more vps on this server next month.

85mb is a really strange usage report. this website was taking 500-600mb on the last host. 1gb-1.2gb on peak hours.

may I send my access details for somebody on the support team to analyse it?

Thanks, and sorry my english.

 

[RenatoMN]

News:

updated proxmox with apt-get update / apt-get upgrade, created a new vm. vzctl is fixed (no more the physpages != 0 error).

btw, the memory usage still seems strange. does a huge forum (200-400 active users all the time) really use only 68MB? i'm worried...

it still give me Internal Server Error sometimes, what make me feel something is wrong

 

[dietmar]

Please use

# ps auxww

to analyze what processes are running and how much memory they use (instead of top).

- Dietmar

 

[RenatoMN]

It's so strange.

outside the container, top shows all memory is in use (and some KBs of swap). ps auxww gave me a looong list of 0.0% MEM use processes (only lighttpd, mysql, and named show 0.1 to 0.4)

inside VM (vzctl enter VMID) top show 68MB use, and ps auxww, a lot of 0.0 (only the same above are using some memory).

It is acceptable, maybe PVE cache all the memory, or whatever.. but it's strange to a huge site, like I said you, use only 68MB memory, and return Internal Server Error sometimes. I just want to be sure its a normal memory use of PVE before blame lxadmin and go complain to lxlabs.

Thanks

 

[tog]

If you login to the proxmox ve host and do "cat /proc/bc/NNN/resources" (where NNN is your vz container ID), do you have any "failcnt" counters that are greater than 0 indicating resource limits have been imposed?

 

[RenatoMN]

            kmemsize                  3515352              6296490  9223372036854775807  9223372036854775807                    0
            lockedpages                     0                    0               491520               491520                    0
            privvmpages                 21149                41162               786432               798932                    0
            shmpages                        0                    0  9223372036854775807  9223372036854775807                    0
            numproc                        33                   63                 1024                 1024                    0
            physpages                   10180                26473                    0  9223372036854775807                    0
            vmguarpages                     0                    0               786432  9223372036854775807                    0
            oomguarpages                10180                26473               786432  9223372036854775807                    0
            numtcpsock                     91                  188  9223372036854775807  9223372036854775807                    0
            numflock                        4                    5  9223372036854775807  9223372036854775807                    0
            numpty                          0                    1                  255                  255                    0
            numsiginfo                      0                    5                 1024                 1024                    0
            tcpsndbuf                 4391424              8602624  9223372036854775807  9223372036854775807                    0
            tcprcvbuf                 1490944              2643200  9223372036854775807  9223372036854775807                    0
            othersockbuf                14848              1266944  9223372036854775807  9223372036854775807                    0
            dgramrcvbuf                     0                17408  9223372036854775807  9223372036854775807                    0
            numothersock                   11                   31  9223372036854775807  9223372036854775807                    0
            dcachesize                 355419               467226  9223372036854775807  9223372036854775807                    0
            numfile                      2144                 3612  9223372036854775807  9223372036854775807                    0
            numiptent                      10                   10  9223372036854775807  9223372036854775807                    0

I don't know which column i should analise here, or if failcnt should be a line here (not present = 0).. please see.

Thanks

 

[tog]

If I understand user beancounters correctly, you haven't yet run into a system/kernel imposed memory resource allocation limit. When you do, you'll know; That last column which is currently all zeros will increment. That is your "failcnt" column.

http://wiki.openvz.org/BC_proc_entries

You can also cat /proc/bc/resources

 

[RenatoMN]

THanks, tog.

I will then assume I should optimize lighttpd or mysql configs.

Thanks all PVE team too, its a wonderful tool

Good week.

 

[tog]

Probably... though I haven't used lighthttpd, I assume you have an error log like Apache does where you can see the output of big nasty errors like could not fork or could not allocate memory?

That'll be your best starting indication of what's going on: The piece of software itself having an error log telling you what happened at the time you got "Internal server error" responses.

If you're using PHP, I hope you're using it with FastCGI and you've provided for an adequate number of fastcgi/php child processes to handle the load. I use PHP with FastCGI on my Apache server because I wanted the hybrid threaded MPM and PHP has all kinds of things plugged into it that aren't thread-safe so can't use mod_php. This way uses far less RAM, too.

When I call PHP via fastcgi I do it with a shell script that can set environment variables so I can configure fastcgi's behavior:

#!/bin/sh

PHPRC="/usr/local/etc/php"
export PHPRC
PHP_FCGI_CHILDREN=4
export PHP_FCGI_CHILDREN
exec /usr/local/bin/php-cgi

You'll want a lot more than 4 if you've got some heavy traffic. Maybe like 20.

 

[RenatoMN]

I use lxadmin, so all things came pre-configurated, i just switch from apache to lighttpd using the panel. I've seen a "fastcgi max instances" or something like it when creating the user that handle the website domain, and it is by default unlimited. I think it use a config like yours.

I will search the logs for errors. Thanks for your info.

Do you know if that BC proc entries, are reseted on VM restart, or on PVE machine restart?

Thanks

 

[tog]

Do you know if that BC proc entries, are reseted on VM restart, or on PVE machine restart?

Thanks

User beancounters are reset on VM restart I believe.