Help with ippool-dhcp servers (two in redundancy)
Hello, I got into a problem and I am not been able to find a solution. I have a scenario where I have two Freeradius IPPool DHCP servers that serve the same network. When an equipment send an request for DHCP both servers get the same request and one of them reply to it. I want to make a delayed response in one of them so I can keep the IP allocation more reliable. I am getting deadlocks in MySQL when running with two servers and I think this is caused by the multiple request in those servers. Atenciosamente, *Nataniel Klug* | nataniel.klug@gmail.com
I got into a problem and I am not been able to find a solution. I have a scenario where I have two Freeradius IPPool DHCP servers that serve the same network. When an equipment send an request for DHCP both servers get the same request and one of them reply to it. I want to make a delayed
I am not totally sure but as I understand it both servers should send a reply, but the client should use the one which replied first. In which case it should not be a problem to have two servers linked to the same backend DB.
On Oct 30, 2017, at 10:34 AM, Nataniel Klug <nataniel.klug@gmail.com> wrote:
I got into a problem and I am not been able to find a solution. I have a scenario where I have two Freeradius IPPool DHCP servers that serve the same network. When an equipment send an request for DHCP both servers get the same request and one of them reply to it. I want to make a delayed response in one of them so I can keep the IP allocation more reliable.
Unfortunately, v3 doesn't really support that kind of delay. v4 will, tho. :(
I am getting deadlocks in MySQL when running with two servers and I think this is caused by the multiple request in those servers.
The better solution is to fix MySQL and the queries. In v3, the queries have a "LIMIT 1" clause, which should help. By default, InnoDB locks the whole row (or really the range being SELECTed) for updates. This is bad. Better DBs like PostgreSQL don't do that. If you're using v3.0.15 with the up-to-date queries, this problem shouldn't happen. Alan DeKok.
Hello Alan and Thompson, @Thompson, I would like just one of the servers to reply and if it's offline the other replies. @Alan, I am running: *radiusd: FreeRADIUS Version 3.0.12, for host x86_64-unknown-linux-gnu, built on Feb 2 2017 at 15:00:22* *FreeRADIUS Version 3.0.12* I don't think I have the up-to-date queries. I would love to run it over PostgreSQL and I know that it is better but I don't have a reliable HA environment for PostgreSQL. I downloaded version 3.0.15 and the queries look exactly the same. I think until version 4 I will leave one of the servers in standby and watching for the other server and, if the first get down the script will enable the second one. Hot standby. Atenciosamente, *Nataniel Klug* | nataniel.klug@gmail.com 2017-10-30 12:34 GMT-02:00 Nataniel Klug <nataniel.klug@gmail.com>:
Hello,
I got into a problem and I am not been able to find a solution. I have a scenario where I have two Freeradius IPPool DHCP servers that serve the same network. When an equipment send an request for DHCP both servers get the same request and one of them reply to it. I want to make a delayed response in one of them so I can keep the IP allocation more reliable. I am getting deadlocks in MySQL when running with two servers and I think this is caused by the multiple request in those servers.
Atenciosamente,
*Nataniel Klug* | nataniel.klug@gmail.com
On Oct 30, 2017, at 12:26 PM, Nataniel Klug <nataniel.klug@gmail.com> wrote:
I would like just one of the servers to reply and if it's offline the other replies.
That's a lot more complicated to configure. FreeRADIUS is a stand-alone server, and doesn't really have provisions for interconnecting multiple servers.
*radiusd: FreeRADIUS Version 3.0.12, for host x86_64-unknown-linux-gnu, built on Feb 2 2017 at 15:00:22* *FreeRADIUS Version 3.0.12*
That should be fine, I think.
I don't think I have the up-to-date queries.
It's easy enough to check.
I would love to run it over PostgreSQL and I know that it is better but I don't have a reliable HA environment for PostgreSQL. I downloaded version 3.0.15 and the queries look exactly the same.
So then it should work. Except that InnoDB is terrible. :( The other option is to double-check the indexes. The default schema indexes the fields in the WHERE clause, which means that it *shouldn't* do table-level locks.
I think until version 4 I will leave one of the servers in standby and watching for the other server and, if the first get down the script will enable the second one. Hot standby.
That's simple. Alan DeKok.
participants (3)
-
Alan DeKok -
Nataniel Klug -
Бенджамин Томпсон