- Integrated Systems
- About Us
- Integrated Systems
- About Us
If you have bookmarked forums or discussion boards in Servers and Operating Systems, we suggest you check and update them as needed.
08-02-2001 03:26 AM
I am currenty looking at limits for expandability on a N class running 11.00.
Max limits can be set in kernel, but what are the max number of telnet sessions, ftp sessions etc that a fully loaded (8 x 550 Mhtz and 32 Gb RAM) can run.
Basically things like 2GB on files can cause problems for tar etc, but what about pttys and systems that control the actual connectivity to the server.
Solved! Go to Solution.
08-02-2001 04:15 AM
To configure your box for large users you must configure
maxproc , maxuproc,
Maxusers is used to set some of these values, so increase first this parameter.
08-02-2001 05:33 AM
Thanks for the reply: -
What I am looking at is things like if I try to telnet 10,000 times to the server at what point will it stop and say cannot allocate specific recourses, for example it will not allow any more telnetd to run of allocate ptty.
I think I should have rephrased my question better.
08-02-2001 10:46 AMSolution
I know my response was very sort, but it was a fast response, and 2 it was the first response.
For 10.000 telnet sessions you should configure npty as 10.000
But 10.000 telnets needs 10.000 process plus 10.000 sh -> at least maxproc = 20.000
Each process use to open 3 standard files -> nfiles=60.000
And so on.
If I recall well sysdef command list actual ,min and max values accepted by kernel.
My preference is SAM, to tunnig.
08-08-2001 04:10 AM
in addition to what Carlos wrote, accessing your system
over the network means usually BSD pseudo ttys, plus
System V pseudo ttys, plus TELNET pseudo ttys, and for
each you will need a master and a slave device.
So we are talking about about a maximum of about
65000 TCP connections concurrently, which would need
65000 * 3 * 2 inodes for those device in your root-fs!
I do remember a customer having a setup like this,
and they had to install specifically for these conditions
(well, that was in the days of HFS for root-fs). Still this
needs an awful lot of inodes, file-entries, processes,
timers (ncallout) and such in your kernel, so the next
step will be to install the patch to boot "large kernels",
which is needed at a certain size of your kernel (I do
remember 79MB for the V-classes, for the others it was
way smaller, something like ~32MB).
And you will need many, many LAN interface cards, to
spread the burden, perhaps even APA (port trunking).
Ask your hp-supporters for more ramifications, as there
are propably many more :-(
04-23-2002 06:06 AM
This (in conjunction with some other parameters) sets the maximum amount of swap space configurable on the system. Maxswapchunks should be set to support sufficient swap space to accommodate all swap anticipated. Also remember, swap space, once configured, is made available for paging (at boot) by specifying it in the file /etc/fstab (/etc/checklist on 9.X). The maximum swap space limit calculated in bytes is: (maxswapchunks * swchunk * DEV_BSIZE). We recommend this parameter be set to 4096.
Hope this helps