Re: [cobalt-users] raq4m acceptable ram usage

[Dmitry Alexeyev <dmi_a@xxxxxxxxxx>]

Hello! 

> Hi Dmitry, thank you very much for your help . As the end of this
> mail i paste my uptime with the memory data as you suggested ( now
> the server is lightly less busy than yesterday ).
>
> is there a tool or something i can use to detect problems in hard
> disk ? i do not have phisical access to the server.
> It would be nice to predict that hd crashes so i can order a new hd.
> I think i will put some more RAM and will monitor swap. thanks.

Predicting hdd crashes is not easy. Well, all I can say that you can 
expect hdd starting to crash in next 3 years ;-) Modern hard disks have 
S.M.A.R.T. system, which does prediction ;) and disables bad clusters. 
There are some smart monitoring tools, no idea how they will work with 
RaQ's ancient kernel, but I'll give them a try; they are now in my TODO 
list (talking about smartmontools, etc - anybody tried them on RaQs? I 
have a desktop PC here, which has exact copy of RaQ software, but 
kernel + initscripts )

If you see something like I described: no services are started, but 
connections are made (easy to check with telnet to port 80 or any other 
open, telnet says connected, but no reply from server is printed), then 
you have to do something. It includes: buying memory, taking care of 
good backup, moving swap partition to some place (it is done with 
parted, I have rpm package for RaQ3/4) or buying a new hdd. 
if hdd has bad blocks, it means SMART system has no more resources to 
work these things around...

I have never seen hard disks crashing unexpectedly, they all were giving 
signs of poor health for several months. 

Best Regards,
Dmitry

>
>
> [admin admin]$ uptime
>   3:58am  up 7 days, 15:52,  1 user,  load average: 2.19, 1.57, 1.55