[SOLVED] worker nodes want to boot faster from the server (!)
Linux - ServerThis forum is for the discussion of Linux Software used in a server related context.
Notices
Welcome to LinuxQuestions.org, a friendly and active Linux Community.
You are currently viewing LQ as a guest. By joining our community you will have the ability to post topics, receive our newsletter, use the advanced search, subscribe to threads and access many other special features. Registration is quick, simple and absolutely free. Join our community today!
Note that registered members see fewer ads, and ContentLink is completely disabled once you log in.
If you have any problems with the registration process or your account login, please contact us. If you need to reset your password, click here.
Having a problem logging in? Please visit this page to clear all LQ-related cookies.
Get a virtual cloud desktop with the Linux distro that you want in less than five minutes with Shells! With over 10 pre-installed distros to choose from, the worry-free installation life is here! Whether you are a digital nomad or just looking for flexibility, Shells can put your Linux machine on the device that you want to use.
Exclusive for LQ members, get up to 45% off per month. Click here for more info.
worker nodes want to boot faster from the server (!)
we have setup a diskless cluster where there is a server which run DHCP and there are some clients that receive an ip address from the server upon the boot.
The problem is, if the nodes (server + clients) restart unexpectedly (power plant problems), then the server will start dhcp service with delay because the fsck is running. At the mean time, the clients quickly pass the POST process and they want to get IP address from server. The server still has not started the DHCP service, so the clients stuck at the first step of network boot.
My questions:
1) is there any way to prioritize the services on server? For example, I want the DHCP to be the first service!!
2) Currently the clients wait for about 10 seconds to get an IP from server. I mean the | / - | \ progress bar. Is there any way to increase this short timeout?
fsck runs at a lower level than a service like dhcp and MUCH earlier in the boot process ie specifically during the disk mount process.
You'll have to wait.
What about a network controlled power outlet for the nodes. Hence, after the server is up, you can switch on the nodes from the server.
NB: Do you have any message on the monitor for the BIOS of the network card? In this menu for the network card, you could try to adjust to reboot if there is no response instead of falling to the next boot option (this is not possible with all cards though).
fsck runs at a lower level than a service like dhcp and MUCH earlier in the boot process ie specifically during the disk mount process.
You'll have to wait.
What about changing the S numbers in /etc/rcX.d/ to change the priority?
Quote:
NB: Do you have any message on the monitor for the BIOS of the network card? In this menu for the network card, you could try to adjust to reboot if there is no response instead of falling to the next boot option (this is not possible with all cards though).
Your clients have to wait; you need to switch your clients on once the server is up (as indicated by Reuti). I don't expect that the startup sequence will make a serious difference for your problem.
You need to read up on the boot process, as your suggestion in post#4 makes no sense.
fsck runs BEFORE the disks are mounted; no way you can run a service until the code it depends upon is available from the disk being mounted.
Currently all nodes (clients and server) are plugged in power source. So in case of power problems and recovery, all nodes start at the same time.
What I understand from your replies, is that adding an external device to delay the power current for the clients. Right?
Something like this:
power source -> server
power source -> delay circuit -> clients
And the nicest solution would be if you don't use a delay circuit but a device controlled by the server (serial port, network); once the server has brought up the dhcp daemon, it can start a process to control that device to power on the clients.
Last edited by Wim Sturkenboom; 08-02-2012 at 04:22 AM.
LinuxQuestions.org is looking for people interested in writing
Editorials, Articles, Reviews, and more. If you'd like to contribute
content, let us know.