Hello all,
Has anyone some experience running RT on top of MySQL Cluster?
Feasible? Overkill? Any issues?
We think that with our application we will see peaks/burts of tickets
submissions. We want to be able to absorb these peaks safely.
Cheers,
Thierry
Hello all,
Has anyone some experience running RT on top of MySQL Cluster?
Do you have experience with MySQL Cluster with other software. I doubt
there is a need for MySQL cluster with RT and I’m sure there will be
issue because unless something has changed, all data must fit in servers
memory when using MySQL cluster and an RT DB with many tickets is a DB
of several GB 
Feasible? Overkill? Any issues?
We think that with our application we will see peaks/burts of tickets
submissions. We want to be able to absorb these peaks safely.
can you tell us more about yout peaks, I know many big setups that just
works fine with one DB. With RT, ticket submissions are often not an
issue, but having a lot of queues/groups/privileged users may slow down
your RT.
Hello all,
Has anyone some experience running RT on top of MySQL Cluster?
Feasible? Overkill? Any issues?We think that with our application we will see peaks/burts of tickets
submissions. We want to be able to absorb these peaks safely.Cheers,
Thierry
Are you sure the bottle-neck is the database backend. It seems
to me that most of the processing is in the webserver front-ends.
If that is the case, point a second front-end to the same backend
database and you will be able to handle more tickets. It is worth
testing, but is your database I/O even close to 100%?
Cheers,
Ken
Hi Thierry,
you should tell us more about your setup, Hardware Sizing and so on. How
many (concurrent) users etc.
Don’t think, that you will solve your problems with a mysql cluster…
Torsten2010/11/19 Thierry Thelliez thierry.thelliez.tech@gmail.com
Hello all,
Has anyone some experience running RT on top of MySQL Cluster?
Feasible? Overkill? Any issues?We think that with our application we will see peaks/burts of tickets
submissions. We want to be able to absorb these peaks safely.Cheers,
Thierry
MFG
Torsten Brumm
Hi Thierry,
I successfull experience RT cluster on Mysql cluster
All works fine on debian lenny with heartbeat (Linux HA) + DRBD (raid 1 over IP - block synchronization).
The /var/lib/mysql is installed on DRBD partition
Apache and RT config are setup identically on both cluster node.
Regards
TanguyLe 19 nov. 2010 à 19:56, “Torsten Brumm” <torsten.brumm@googlemail.commailto:torsten.brumm@googlemail.com> a écrit :
Hi Thierry,
you should tell us more about your setup, Hardware Sizing and so on. How many (concurrent) users etc.
Don’t think, that you will solve your problems with a mysql cluster…
Torsten
2010/11/19 Thierry Thelliez <mailto:thierry.thelliez.tech@gmail.comthierry.thelliez.tech@gmail.commailto:thierry.thelliez.tech@gmail.com>
Hello all,
Has anyone some experience running RT on top of MySQL Cluster?
Feasible? Overkill? Any issues?
We think that with our application we will see peaks/burts of tickets
submissions. We want to be able to absorb these peaks safely.
Cheers,
Thierry
MFG
Torsten Brumm
http://www.brumm.mehttp://www.brumm.me
http://www.elektrofeld.dehttp://www.elektrofeld.de
Thanks all for the quick answers.
In terms of sizing, we could have several thousands (max = 10,000;
more realistically 3,000) tickets submissions within a day. They will
not be nicely spread during the day. Some hours might be a lot busier
than others. Each ticket will have at least one attachment (about 1 to
3 mb each).
I am just starting to learn about MySQL Cluster, trying to find an
architecture that could be robust enough to minimize data loss in case
of failure. In other words, we do not want to lose hours of submitted
tickets if something bad happens. Likewise, we want to provide
continuity of operations even if a server goes down.
From what I read in this forum, RT should be fine with the volume
assuming a decent server. My concern is more about robustness.
Thanks for your comments and suggestions.
Thierry
And to respond to Emmanuel’s question about queues/group/privileged
users, this is a simple setup… 2 queues, 5 groups, and a dozen
privileged users and few hundred unprivileged ones. So I understand
that this is ok?
Thierry Thelliez
You can always just use MySQL replication to keep a hot backup of the DB
on another server/site. This is what we do, and it works out pretty well.
Failover isn’t automated in any fashion, but in theory would just be a
config change to point to the new DB server.
-EricFrom: rt-users-bounces@lists.bestpractical.com
[mailto:rt-users-bounces@lists.bestpractical.com] On Behalf Of Thierry
Thelliez
Sent: Friday, November 19, 2010 4:16 PM
To: rt-users@lists.bestpractical.com
Subject: Re: [rt-users] RT and mySQL cluster?
Thanks all for the quick answers.
In terms of sizing, we could have several thousands (max = 10,000; more
realistically 3,000) tickets submissions within a day. They will not be
nicely spread during the day. Some hours might be a lot busier than
others. Each ticket will have at least one attachment (about 1 to
3 mb each).
I am just starting to learn about MySQL Cluster, trying to find an
architecture that could be robust enough to minimize data loss in case of
failure. In other words, we do not want to lose hours of submitted tickets
if something bad happens. Likewise, we want to provide continuity of
operations even if a server goes down.
From what I read in this forum, RT should be fine with the volume assuming
a decent server. My concern is more about robustness.
Thanks for your comments and suggestions.
Thierry
Hi Thierry,
I successfull experience RT cluster on Mysql cluster
All works fine on debian lenny with heartbeat (Linux HA) + DRBD (raid 1 over IP - block synchronization).The /var/lib/mysql is installed on DRBD partition
Apache and RT config are setup identically on both cluster node.
speaking of DB, you’re not using the “cluster” mode of MySQL, you have
just a failover between those two servers which is IMHO a good setup for
failover
Thanks all for the quick answers.
In terms of sizing, we could have several thousands (max = 10,000;
more realistically 3,000) tickets submissions within a day. They will
not be nicely spread during the day. Some hours might be a lot busier
than others. Each ticket will have at least one attachment (about 1 to
3 mb each).
you have to do benchmark, but here “cluster” mode of MySQL will not help
at all (to my knowledge it may decrease performances).
here I will go with good hardware (RAID 10) and one DB.
I am just starting to learn about MySQL Cluster, trying to find an
architecture that could be robust enough to minimize data loss in case
of failure. In other words, we do not want to lose hours of submitted
tickets if something bad happens. Likewise, we want to provide
continuity of operations even if a server goes down.
using RT on a failover architecture is pretty easy and a good idea. You
just have to setup either a master/slave MySQL configuration, or drbd,
or a hot backup MySQL.
And to respond to Emmanuel’s question about queues/group/privileged
users, this is a simple setup… 2 queues, 5 groups, and a dozen
privileged users and few hundred unprivileged ones. So I understand
that this is ok?
yes
Sounds like a Overkill for your setup. Think MySQL Master/Slave is what you are looking for like Emmanuel Said.
TorstenAm 19.11.2010 um 22:27 schrieb Thierry Thelliez thierry.thelliez.tech@gmail.com:
And to respond to Emmanuel’s question about queues/group/privileged
users, this is a simple setup… 2 queues, 5 groups, and a dozen
privileged users and few hundred unprivileged ones. So I understand
that this is ok?Thierry Thelliez