We have a BizTalk Server install backed by a clustered SQL environment for high availability.
However, whenever the SQL environment is patched there is a momentary outage as part of node failover. Consequently the host instances stop and BizTalk shuts down (if we move to CU2 the host instances will automatically restart, but this is a separate issue).
This is undesirable, as it prevents incoming web requests and breaks open web service clients. As such, is there a strategy for gracefully patching SQL Server without a BizTalk outage?
It seems this is impossible. Marking this as the accepted answer until someone can pleasantly surprise me otherwise.
Related
I am working in a project where we are using Go as a web server and MySQL.
We have been told to implement fault tolerance to handle a hardware crash. We were given 2 servers which have MySQL and the Go-server on them.
We have succesfully set up replication in MySQL, but we are struggling with the failover part. Our thought was to get an extra server with HAProxy to have a primary server and then being able to failover to the backup server.
We also considered using MySQL failover, but did not see how we could redirect the traffic using it.
Is this a reasonable plan? Or what would you recommend that we do instead?
If you want two identical servers connecting to their local MySQL instances, you need a way of deciding which one is the production server. There are a number of solutions for that, including
Setting up a reverse proxy, as you mention, but then, your proxy
itself becomes a SPOF,
Using a floating IP, also known as a failover
IP, but this only works if your host supports it. Cloud providers
typically support them, as well as some bare metal server providers.
There is nothing specific to Go as far as I know.
I have a spark 1.2.1 cluster set up in standalone mode with a master and a few slaves. I then let my data scientists enjoy the cluster's power.
All is working fine. However, the dedicated server that my data scientists used to submit spark jobs have its spark.local.dir filled up gradually.
Given that this machine is sitting outside of the cluster, not a master, nor a worker/slave, I wouldn't think that the local spark.local.dir is used in any way by spark. (And why would it? It only shows the logs.)
I could not find a good doc detailing this part of information. Does anybody have an idea?
Not enough information about your setup to be sure, but I am guessing that the jobs are launched in client mode where the driver would be on your client node.
From the spark docs:
In client mode, the driver is launched in the same process as the client that submits the application. In cluster mode, however, the driver is launched from one of the Worker processes inside the cluster, and the client process exits as soon as it fulfills its responsibility of submitting the application without waiting for the application to finish.
I am guessing that in client mode the driver (on your client machine) of the application needs plenty of scratch space to manage the other workers in that case.
I have been running a MySQL Community Server for a couple of years now and a new client has asked for a report from a vulnerability scanner on our network. I am using OpenVAS and the network is fine apart from the server, its returning a high threat stating that a MySQL security patch needs to be applied. I've gone onto the Oracle website and I believe that I require a Support Identifier to apply the patch, so I done some Googling and its basically a subscription from Oracle. As its a small company is there a way to apply this patch for the community edition without the need to fork out a ton of money, or shall I just filter incoming traffic to the mysql port (Its not the actual fix but at least its one)?
Cheers for the help!
A first measure would be closing the MySQL port through a firewall (iptables), or at least restricting it to the machines in the internal network needing direct access to MySQL.
As for the patch: Maybe there are newer pre-built packages for your OS/distro which already contain the bugfix.
I am currently running a virtualized environment for my web and db server. When I access the web server or the MySQL server individually, they are both fast. I also have websites running on the web server that do not require the db server and those all load quickly. However, when I access my hosted website that requires the web server to call from the db server, there is about a 5-7 second latency for every page load. This has been confirmed with both a very simple site and with a Word Press setup as well. Here is the config:
Web server - CentOS 6.5, Apache 2.2.15
DB server - CentOS 6.5, MySQL 5.1.73
My question is, are the servers continuously authenticating with one another (and thus causing latency) on every single db call? If that is the case, does anyone know how to permanently authenticate between the two?
I might be way off on this assumption and authentication could have nothing to do with it. I am completely open to any and all ideas at this point. Thank you very much.
V/R,
Tony
To me it seems to be a network issue.
and obviously the db-server will need authentication every time there is a hit.
For our customer the application which is running is using MySQL database. However, this server is without monitoring. I want to install OpenNMS (which uses PostgreSQL) application to monitor the solution and send the traps to main NMS system.
Is there any problem having both on the same server?
No, there is no technical problem. Both default to different ports they listen on.
The only problem that could arise is that each individual DB might be slower compared to an installation on separate phyiscal machines because they are both share (and fight for) for the same resources (I/O, memory, CPU, network, ...)