Server Uses Too Much RAM

[HighCommander540]

I have been running a Ubuntu 7.10 LAMP server for about a year now. Up until a few weeks ago it has been using very little RAM, like around 100 MB (I don't have the most popular site).

Recently it has been using increasingly more RAM, when I check usually around 300MB. There hasn't been anymore traffic...I first thought it was the BASH scripts I was using, because I just learned how to write them a couple weeks ago and that is around the time of the increase, but I stopped all of them and still the increase.

I have noticed that a lot more instances of Apache are running now than I have ever seen before. IDK, why.

Also, I do run a TeamSpeak server, but they is not much activity there either. I do have it set to autostart with a cron job, that uses a BASh script. But the cron job always zombies for some reason and shows up as defunct.

Can someone please help me figure out what the problem is?

[Cracauer]

How do you measure how much RAM is used by what?

[linux_tech]

How do you measure how much RAM is used by what?

In terminal type:


free -m
top


to get a est of memory usage

[HighCommander540]

In terminal type:


free -m
top


to get a est of memory usage

Yes, that is exactly how I check the RAM usage. First, I usually run a "free -m" to see how much is being used and then I use "top" to find out what is using the most.

Usually, top -u www-data. And I find that there are a lot of Apache threads running.

Anyone have any advice?

[Cracauer]

That's not useful to debug performance problems. Unix/Linux will always take up all RAM, in fact it is a problem when it doesn't.

[HighCommander540]

No, that is wrong...Very wrong. The system will cache all of the RAM, but it knows how much of the actual RAM it is using for processes and not just in its cache. MAC OS X does the same thing and both know how much their processes are actually using.

If your computer tells you that itis "using" all of your RAM all the time, then there is something considerably wrong with your computer sir.

So is there anyone that has any actual advice? Because it didn't always use this much RAM.

[Cracauer]

So your claim is that your "server" (I assume that means apache) takes up more RAM now and that interactive processes are slowed down by lack of available memory now?

Can you post, exactly, how much RSS your interactive programs had before and after?

[cariboo907]

I would be really helpful if you pasted the outputs of:

free -m

and

top

in your next post. Just ssh into your server and run the apps and paste the output from your terminal into your post. Remember to use the tags.

Jim

[HighCommander540]

So your claim is that your "server" (I assume that means apache) takes up more RAM now and that interactive processes are slowed down by lack of available memory now?

Can you post, exactly, how much RSS your interactive programs had before and after?

No, they don't run any slower and nothing at all on the system is affected. No speed difference, its just that there is more RAM being used now then ever before for no apparent reason.


free -m:

total used free shared buffers cached
Mem: 3034 408 2626 0 139 163
-/+ buffers/cache: 105 2929
Swap: 3153 0 3153

top:

top - 04:25:19 up 6 days, 23:33, 1 user, load average: 0.00, 0.00, 0.00
Tasks: 59 total, 2 running, 57 sleeping, 0 stopped, 0 zombie
Cpu(s): 0.0%us, 0.0%sy, 0.0%ni,100.0%id, 0.0%wa, 0.0%hi, 0.0%si, 0.0%st
Mem: 3107836k total, 418040k used, 2689796k free, 142784k buffers
Swap: 3229024k total, 0k used, 3229024k free, 167452k cached

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
20548 zac 20 0 2308 1084 856 R 0.3 0.0 0:00.04 top
1 root 20 0 2844 1692 548 S 0.0 0.1 0:01.61 init
2 root 15 -5 0 0 0 S 0.0 0.0 0:00.00 kthreadd
3 root RT -5 0 0 0 S 0.0 0.0 0:00.00 migration/0
4 root 15 -5 0 0 0 S 0.0 0.0 0:00.04 ksoftirqd/0
5 root RT -5 0 0 0 S 0.0 0.0 0:00.00 watchdog/0
6 root 15 -5 0 0 0 S 0.0 0.0 0:03.11 events/0
7 root 15 -5 0 0 0 S 0.0 0.0 0:00.00 khelper
41 root 15 -5 0 0 0 S 0.0 0.0 0:00.51 kblockd/0
44 root 15 -5 0 0 0 S 0.0 0.0 0:00.00 kacpid
45 root 15 -5 0 0 0 S 0.0 0.0 0:00.00 kacpi_notify
116 root 15 -5 0 0 0 S 0.0 0.0 0:00.00 kseriod
150 root 20 0 0 0 0 S 0.0 0.0 0:00.00 pdflush
151 root 20 0 0 0 0 S 0.0 0.0 0:02.05 pdflush
152 root 15 -5 0 0 0 S 0.0 0.0 0:00.00 kswapd0
194 root 15 -5 0 0 0 S 0.0 0.0 0:00.00 aio/0
1321 root 15 -5 0 0 0 S 0.0 0.0 0:00.00 ksuspend_usbd
1325 root 15 -5 0 0 0 S 0.0 0.0 0:00.00 khubd
1353 root 15 -5 0 0 0 S 0.0 0.0 0:00.00 ata/0
1355 root 15 -5 0 0 0 S 0.0 0.0 0:00.00 ata_aux
1581 root 15 -5 0 0 0 S 0.0 0.0 0:00.00 scsi_eh_0
1583 root 15 -5 0 0 0 S 0.0 0.0 0:00.00 scsi_eh_1
2097 root 15 -5 0 0 0 S 0.0 0.0 0:00.01 scsi_eh_2
2098 root 15 -5 0 0 0 S 0.0 0.0 0:00.01 scsi_eh_3
2317 root 15 -5 0 0 0 S 0.0 0.0 0:15.47 kjournald
2470 root 16 -4 2224 672 380 S 0.0 0.0 0:00.47 udevd
2909 root 15 -5 0 0 0 S 0.0 0.0 0:00.00 kpsmoused
3756 dhcp 18 -2 2440 772 452 S 0.0 0.0 0:00.01 dhclient3
4055 root 20 0 1716 516 444 S 0.0 0.0 0:00.00 getty
4056 root 20 0 1716 516 444 S 0.0 0.0 0:00.00 getty
4060 root 20 0 1716 512 444 S 0.0 0.0 0:00.00 getty
4061 root 20 0 1716 512 444 S 0.0 0.0 0:00.00 getty
4063 root 20 0 1716 508 444 S 0.0 0.0 0:00.00 getty
4104 syslog 20 0 1936 684 532 S 0.0 0.0 0:07.49 syslogd
4123 root 20 0 1872 544 448 S 0.0 0.0 0:00.04 dd
4125 klog 20 0 3152 2016 424 S 0.0 0.1 0:00.11 klogd
4144 root 20 0 5312 1024 672 S 0.0 0.0 0:02.51 sshd
4328 root 20 0 3628 956 748 S 0.0 0.0 0:00.67 vsftpd
4353 daemon 20 0 1984 420 300 S 0.0 0.0 0:00.00 atd
4364 root 20 0 2104 892 712 S 0.0 0.0 0:00.50 cron
4424 root 20 0 1716 516 444 S 0.0 0.0 0:00.00 getty
19287 root 20 0 23276 6788 3904 S 0.0 0.2 0:00.85 apache2
19293 www-data 20 0 27068 10m 4192 S 0.0 0.4 0:01.70 apache2
19338 www-data 20 0 28360 12m 4184 S 0.0 0.4 0:01.61 apache2
19360 www-data 39 19 95900 2056 1452 S 0.0 0.1 0:42.60 server_linux
19458 root 20 0 1772 524 440 S 0.0 0.0 0:00.00 mysqld_safe
19497 mysql 20 0 125m 17m 5092 S 0.0 0.6 0:13.13 mysqld
19499 root 20 0 2920 696 596 S 0.0 0.0 0:00.00 logger
19570 www-data 20 0 26596 10m 4200 S 0.0 0.3 0:01.51 apache2
19658 www-data 20 0 26832 10m 4252 S 0.0 0.3 0:00.96 apache2
19659 www-data 20 0 27856 11m 4352 S 0.0 0.4 0:01.10 apache2
19660 www-data 20 0 26788 10m 4100 S 0.0 0.3 0:00.74 apache2
19898 www-data 20 0 26580 10m 3960 S 0.0 0.3 0:00.36 apache2
19899 www-data 20 0 26828 10m 3660 S 0.0 0.3 0:00.59 apache2
20093 www-data 20 0 31440 15m 4308 S 0.0 0.5 0:01.43 apache2

That is what I can get on the screen. The rest of the processes are off the window.

[Cracauer]

You are making up a problem you don't have.

This is all perfectly working.

[Thirtysixway]

Doesn't matter if I have vmware running on my server or not, it says that my free memory is "5". Linux just keeps a hold of your ram and give it to applications.

[HighCommander540]

Ok, well I will take your word on it. I just find it odd that it would do this out of no where. Thanks for your help guys. I apologize for the distress.

[ryanfx]

Don't worry about RAM being taken up. Unused memory = wasted memory.

People complain about OS's like Vista taking up too much RAM because it preloads commonly used apps in in available memory. When the application is called the computer simply retrieves it from memory instead of accessing it from the hard drive first.

[jamessnell]

This seems like a problem I've been having, though my server is running out of memory - it eventually dies with the console showing a bunch of out of memory error messages (if I leave it long enough). My website is also quite low traffic.

Here's output from top -u www-data (this server has an uptime of about 5mins when I ran this command):
top - 16:42:45 up 8 min, 1 user, load average: 0.11, 0.12, 0.08
Tasks: 80 total, 1 running, 79 sleeping, 0 stopped, 0 zombie
Cpu0 : 11.0%us, 1.0%sy, 0.0%ni, 87.7%id, 0.0%wa, 0.0%hi, 0.0%si, 0.3%st
Cpu1 : 0.0%us, 0.0%sy, 0.0%ni,100.0%id, 0.0%wa, 0.0%hi, 0.0%si, 0.0%st
Cpu2 : 0.0%us, 0.0%sy, 0.0%ni,100.0%id, 0.0%wa, 0.0%hi, 0.0%si, 0.0%st
Cpu3 : 0.0%us, 0.0%sy, 0.0%ni,100.0%id, 0.0%wa, 0.0%hi, 0.0%si, 0.0%st
Mem: 368836k total, 321640k used, 47196k free, 2304k buffers
Swap: 1048568k total, 13892k used, 1034676k free, 39152k cached

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
3064 www-data 16 0 201m 34m 4780 S 12 9.6 0:22.68 apache2
3058 www-data 15 0 204m 37m 4772 S 0 10.4 0:03.11 apache2
3059 www-data 15 0 292m 63m 8100 S 0 17.6 0:00.89 apache2
3060 www-data 15 0 179m 10m 3036 S 0 2.9 0:00.02 apache2
3061 www-data 15 0 209m 42m 4764 S 0 11.8 0:00.36 apache2
3062 www-data 18 0 178m 8636 1780 S 0 2.3 0:00.00 apache2
3065 www-data 15 0 204m 37m 4752 S 0 10.4 0:00.41 apache2
3066 www-data 15 0 179m 10m 3044 S 0 2.8 0:00.00 apache2
3067 www-data 15 0 178m 10m 3036 S 0 2.8 0:00.00 apache2
3068 www-data 15 0 210m 42m 4752 S 0 11.9 0:00.30 apache2

When I run free -m it eventually will show 0 free memory in swap and like 4mb for ram.

Any suggestions on what to consider next?

[Cracauer]

Looks fine so far. Why don't you post the same top output from right before the crash?

And don't only include www-data. In fact `ps -alxwwww` would be more useful.

[jamessnell]

hmm, I'm less convinced I have a real problem. My swap was set to 512MB before and my box only gets 360MB of ram. I've since increased the swap to 1024MB and I haven't noticed a problem since.. Maybe I just need more than 360+512 for a LAMP setup?

[Cracauer]

hmm, I'm less convinced I have a real problem.

That is what I was thinking.

My swap was set to 512MB before and my box only gets 360MB of ram. I've since increased the swap to 1024MB and I haven't noticed a problem since.. Maybe I just need more than 360+512 for a LAMP setup?

With mysql and php in there? Certainly.

You see we all ran 256-512 MB machines for web applications in 1996 or whereabouts. But all this software is really not prepared to run on that low memory anymore, the kernel in particular.

And even at the time you gave it a GB of swap or so.

[jamessnell]

Yeah, I'm kinda mortified to seem to need that much swap, but whatever..

[Akita]

Do you have a software RAID array? If your memory "problems" happen once a month and you have software RAID, look for an mdadm job in /etc/cron.d. By default there's a once per month job that does a consistency check on RAID arrays, and it does the damn thing in RAM. I watched a system of mine use all of it's swap and try to die once a month for 3 months after upgrading a RAID1 array to a pair of TB drives. The check every month used 100% RAM AND the swap. Got rid of the cron job and the once-per-month problem went away .. go figure. I think that default check was fine in the days of 20GB drives - LOL.

[jamessnell]

Interesting notion, but no, I don't have a software raid setup. I WAS noticing this problem very frequently - as in within a few mins of rebooting the machine.

Since I doubled the swap space earlier today I still haven't seen the box return to that insane sort of use as before. I think all it really was, was something "lame" in Apache and MySQL effectively requiring a certain amount of stuff to be cached and my machine not initially having the total memory to handle it all (I mean memory = ram + swap).

I guess it's all good now. I'll post more if I see a nasty return to how it was.. But given how fast it happened before and that it's been quite a few hours - I think I *may* be out of those particular woods.

Thanks for the support friends,

Cheerio,
J