Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

MentalFabric

macrumors 6502
Original poster
Mar 10, 2004
373
25
Hi All,

Apologies in advance for the wordiness of my problem!:

I'm working in a publishing company running about 35 mac/5 windows machines.

We have a 10.4.11 Communigate Pro mail server running fine with nothing else but backup

and our 10.5 main server (Running OD, AFP, SMB, DNS and Software Update) connected to a PROMIS RAID which has less than 1Tb of data on it.

Both are quad core Mac Pros, with 2.66Ghz/2.8Ghz 5gb/4gb RAM respectively

On the OD Server, all the services run fine, but there are very frequent lags/stalls in performance of simple tasks such as moving a finder window (can sometimes take 20 mins for the windows to visibly move, at which point everything starts running normally again until the next stall)

Occasionally, all the services will encounter an error (if Server Admin is open) and AFP at least ceases to function for a minute or so.

These symptoms are far more likely to occur when running a backup through Carbon Copy Cloner or Chronosync (usually a few hours in to a BIG backup).

I initially suspected that this must have been caused by overloading the system, but in activity monitor I can never find the CPU running at more than 25% capacity, and there always seems to be over 1Gb of inactive RAM (though rarely more than 100mb free RAM, normally about 20mb). Disk activity does seem to spike here and there but doesn't seem to cause any problems.

So my first question is really - is this normal behaviour for a 10.5 Server? I wouldn't have thought so in such a small company...
Everything seems to run well provided I don't touch it. If anything needs changing on the server though, even the smallest jobs are prone to taking 30mins due to waiting for the server to start responding again between tasks.

Hopefully the answer to the first question is no, in which case I'm looking for any advice I can get that's preferably not "re-install the server" :D

Thanks for reading all the way through!

Kind Regards,

John
 
Sounds like not enough RAM. RAM is cheap get more.

Thanks for responding!

Are you confident though? From what I've heard, it's standard for free RAM too be low unless you've just started up, as the inactive RAM keeps as much of what you've stored in the RAM as it can until it needs to give some of that RAM to another app...
 
Thanks for responding!

Are you confident though? From what I've heard, it's standard for free RAM too be low unless you've just started up, as the inactive RAM keeps as much of what you've stored in the RAM as it can until it needs to give some of that RAM to another app...

Look at the page outs number. If it is more than 0 it means you have run out of RAM for short amount of time. If it is a very large number you need more RAM urgently.
 
Look at the page outs number. If it is more than 0 it means you have run out of RAM for short amount of time. If it is a very large number you need more RAM urgently.

VM size 55.28GB
Page ins: 812.72MB
Page outs: 2.16GB
Swap used 1.35GB

So more RAM it is i suppose! Thanks!

Now, as you're here, would you mind possibly explaining what the page in/outs are? ^.^
 
Page-ins = RAM data moved around "IN" RAM >> okay.
Page-outs = RAM data moved around HD!! >> NOT okay. you need more RAM.

You can also take a look at /var/vm:
A swapfile0 (64 MB) is created @ boot, which is default.
A swapfile1 (64 MB) is created whenever swapfile0 is not enough!
A swapfile2 (128 MB) s created whenever swapfile0+1 is not enough!
A swapfile3 (256 MB).. etc.. etc.

So, if you find that with normal use you have swapfiles0-3 listed in /var/vm, you can say you have needed 512 MB of HD space for swapping purposes... i.e. you needed 512 MB or more physical RAM.
 
Thanks guys! I've ordered another 2x2Gb chips for it.

I'm fairly certain it was the backup pushing it too far - before I posted it was made much worse by starting fresh 1Tb backups - it's been running with over 500Mb free RAM for a couple of days now (page ins getting slowly higher, page outs staying the same) - so I'm still worried that even with double the RAM I've got, some flaw in the software will allow it to consume too much RAM to keep the server running stably. Am I paranoid and making things up? :eek:

And finally, is it normal to have various not-particularly-dangerous-looking error messages in the server logs? Failed lstats mostly...
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.