We have a database cluster that uses MySQL on DRBD with Pacemaker+Corosync, which is great. My question is whether it is possible (and whether anyone has been successful) to mount a CephFS drive to /var/lib/mysql and run MySQL from there.
The advantage would be that MySQL could then run on any cluster node, without the 2 node constraint of DRBD (or worse, the complexity of DRBD on DRBD).
I have a pair of pfSense firewall/routers set up in CARP/XML Config cluster. On the LAN side, the switch also has a pair of servers running corosync/pacemaker/drbd. These are on a different ip network, but still generate multicast packets.
For the life of me, I cannot get pfSense to allow the packets. I tried using the easy rule button, but that failed. I also added a rule that allows all ports, all addresses with a destination of the multicast address, and enabled "allowopts" and "nostate"; all to no avail. The traffic is still stopped by the default rule. Any idea what I might be doing wrong?
Here is a shot of the rules (and yes, they've been reloaded a few times:
I've also tried "no state." The rule under the title there is the Easy-Rule, and it chose the 239 address for both the source and destination; the src port is * and the dest port is 5405.
Here is the log showing the rejection by the default rule:
It's worth noting that it originally showed the scrubbing rule was also blocking, so I disabled the packet fragment scrubbing.
I've read up a little bit on bonding nics with ifenslave; what I'm having trouble understanding is whether there is special configuration needed in order to split the bonds across two switches. For example, if I have several servers that all have two nics each, and two separate switches, do I just configure the bonds and plug 1 nic from each into switch #1 and the other from each into switch #2? or is there more to it than that?
If the bonds are active-backup, will a nic failure on single machine mean that server may become disconnected since the rest of the machines are using the primary nic and it's using the secondary? Or do you link the switches with one cable as well?
I've used ext3 and ext4 in the past with no issues, but I've seen a lot of installations on XFS as well. What are the advantages of XFS vs ext4 vs ?? with respect to the /var/lib/mysql directory?
Does the structure and size of your database influence the suitability of each? Are there other considerations?
Also, does filesystem choice impact High Availability at all?
We make extensive use of gitolite, both for code management and to host private SparkleShare repositories. I'm looking at building a FreeNAS storage machine, and got to thinking that it runs bsd, has ssh, and therefore should theoretically have all it needs to run gitolite. Has anyone tried this? Is there any reason it shouldn't work?
I'm looking at implementing a mirrored pair of FreeNAS servers as file repositories. The only question still to be answered is the physical security. I typically encrypt the underlying drives on all of my servers, but it looks like that is a legacy-only feature for FreeNAS, and ZFS doesn't have an open solution to that yet.
Are there other admins with this issue who have figured out acceptable ways to solve it?
I'm running MySQL for most of our DBs right now, and would like to get off the Oracle ship before it takes more of a turn. My understanding is that the transition from MySQL->MariaDB should be trivial, but going the other way is not so easy. Has anyone made this jump yet?
Edit: Bonus points if you have anecdotal information on doing this with Debian/Ubuntu.
I'm working on a network restructure that has three geographical locations, but will share some services. Two of the locations have workstations, and one has only servers (in a CoLo facility). We will be running PfSense firewalls, and several hosts will provide services inside the LAN and through port-forwards to the internet.
It doesn't seem like it would be worth the hassle to run 4 different DNS views in bind, but it sounds like the load and configuration overhead in PfSense to utilize NAT Reflection would be considerable. What are the caveats to each, or is there another option?
I'm running a VM to test a PfSense installation, and would like to be able to test features with the host machine. I found where to change the address explicitly, and to disable VirtualBox's DHCP server, but couldn't figure out how to make it request an address from my guest.
I have a cluster of Apache webservers using round-robin for primitive load balancing. Is there a way to have firefox or chrome request that server name from a specific IP?
If you just enter the http://1.2.3.4 , it doesn't get the ServerName, and if you enter http://servername, you could get any of the servers.
I have a Hobbit server set up with a handful of hosts using conn, http, ssh, and sslcert services, but would like to add the other tests as well. I've installed hobbit-client on a server, and added:
# CLIENT:fqdn.example.com
to it's host line in bb-hosts, and added:
HOST=fqdn.example.com
before the default configuration in hobbit-clinets.cfg, but no joy. Does anyone know what else I need to do for those tests to register?