geoaxis's questions

I would like to install a Test/Dummy SMTP server on a Windows 2008 server (virtual box). I just want to test my web application on the machine it self so I don't need the mails to go out on the internet, but just to be written to disk (so that I can verify that the mail function was indeed called and the correct data was handed over to SMTP)

Can you recommend some tool. I guess starting your own SMTP server in python is an option. I am looking for a simple (ready to use) solution, targeted for tests systems. I will need to integrate it to automated tests (Selenium) at a later stage.

I have configured OpenNMS to collect all kinds of numeric data coming out of tomcat7 jmx. There are a lot of things. I am interested in monitoring this tomcat instance so that I can avoid down time and lockups. What metrics should I be watching out for?

I am already monitoring things like CPU, Memory, Network via SNMP. With this JMX connection the things that I find interesting are

• Catalina:type=GlobalRequestProcessor,name="ajp-bio-/a.b.c.d-XXXX"
  • RequestsCount so far.
• Catalina:type=Manager,context=/myApp,host=localhost
  • Active sessions and its maximum so far

I am trying to monitor detailed memory useage over time for MacOSX /Xserver. NetSNMP version 5.4.2.1 on MacOS 10.6 does not report the following mibs

 oid=".1.3.6.1.4.1.2021.4.11"  alias="memTotalFree" type="integer"
 oid=".1.3.6.1.4.1.2021.4.13" alias="memShared"    type="integer" 
 oid=".1.3.6.1.4.1.2021.4.14"  alias="memBuffer"    type="integer"
 oid=".1.3.6.1.4.1.2021.4.15" alias="memCached"    type="integer" 

This means that I cannot use NetSNMP to monitor memory effectively. If I can get these stats from some command line utility then they can be exposed to SNMP and I can collect them easily, but I don't see any low level MacOS specific command line tools for reading these statistics.

I would appreciate if you can recommend any tool or collector which would help me store memory statitics for MacOS/XServe for a longer period of time.

I have an ssh script which use ssh key forwarding on a remote system to do thing in even more remote systems.

I have to run the script in a screen session so that I can detach from it. But the authorization that was granted to the remote system by ssh-key-forwarding needs to be present. Atleast until the script has finished.

Any ideas how to go about it?

For an application I use JVM Max heap setting when starting it on both a test environment and a production system. When i connect via JMX to monitor the application I see that the max heap size of production app is not the same as test application. I use -Xmx2000m on both and on test environment I get 1,864,171,520 B and on production I get 2,009,726,976 B.

So my questions are -Is this info reliable. The Java versions and the machines are identical (Mac OS X 10.5).

-If this info is reliable then what could be causing this and how can it be changed so both systems behave the same for max heap size setting.

I am interested in setting up JMX monitoring with minimal configuration as possible. In VisualVM or in JConsole a lot of metrics are shown just by connecting to the JMX server. I would like to store those (important ones) and setup an alarm when they cross a particular threshold.

I have looked into OpenNMS, Nagios, Splunk and Zenoss so far. None of them have the out of box experience. At this point I am more interested in getting work done (I am a Dev guy who moonlights in Ops)

Edit: By out of the box I meant, no need for installing plugins (like Splunk or Zabbix). No clunky configurations (like in OpenNMS). Some thing that just works by minimal effort, give in your jmx url and password and baaam!

I would like to know which key was used when logging into an SSH session. I wan to correlate the key to a local database and send email to an address which corresponds to the key.

The first step is to find which key is the one being used.

I would like to monitor a network device (a disarray) which is attached to a server with its own local network (10.0.0.x). The server is reachable from another private space (172.16.x.x).

I can do snmpwalk between my monitoring server and the machine in question. But I would also like to monitor the disk array attached to the server in question.

Some sort of SNMP proxying or NATing comes to mind. So this could be a question on how to enable NAT (which would allow udp) on MacOS

Thanks

I am looking into installing Linux on a test XServe machine and see how it goes overall. We have our entire infrastructure on OSX and would like to move to Linux servers.

Is there any distribution which has good support for XServe? Google search yielded nothing conclusive. I don't mind hacking around for personal use (compiling kernel, tweaking packages and configurations) but when running in production it's good to be cautious.

Paying for commercial support is definitely an option.

I have a bunch of *nix servers in production. I would like to put it's important configuration files in version control

The basic idea I had was to

• Put a distributted version control system so that each system's configurations are saved on its own drive.

• The way configurations are changed are either through an editor or through an external source.

• Assuming that $EDITOR is used in shell, it would be possible to add an extra command upon saving the file (so the file is then added to git). This could be problematic as it is not always that a user would do vim /etc/my.cnf , it could be opening my.cnf from within vim

• The other source of change can be captured via a cron script, which runs say every one hour and adds stuff to version control (this is good enough). Then the git repo can be pulled into a central place for safekeeping. 1 hour is good enough as it is assumed that first way of capturing changes will work most of the time.

I would like to know that if there is some tool which already does this (or does it better). It would be ideal that on each configuration change initiated by a user, s/he is propted to add the change into the local repo.

I have two servers connected on the same subnet with each other. I would like to know which switches come in between the two servers. I do have admin and physical access to the boxes but they are all plugged in and I would like the least intrusive way to know about IPs of the switches in between.

I guess I would have to go a level down than TCP/IP for this, map the ethernet addresses some how to IPs but not really sure.

I am looking into MySQL Enterprise backup solution. I am interested in a non-blocking mysql backup (live and incremental).

MySQL Enterprise provides this solution (we have the subscription already). But from what I see it's only a console application (and not the most user friendly one).

I am not interested in making wrapper scripts or GUI for MySQL Enterprise Backup. Can you recommend some thing which eases the operation of live backup.

Possible Duplicate:
Dropbox alternative

I would like to setup an automated, dropbox like syncing between two machines which both have Bash/SSH/SCP/rsync etc (they could be Mac, Linux or windows with Cygwin). The syncing should ideally be automated (ie some sort of easy to use GUI based options to schedule downloads, or backups whenever a system gets connected to the internet). It would be nice that if the partial files can be copied (ie if a file is being downloaded to sync folder, its partial contents also get synced)

My searches have turned up http://fak3r.com/geek/howto-build-your-own-open-source-dropbox-clone/ which is based on lsyncd. I would be interested to know more options. Thanks

Edit: The sync will be over the internet. Assume open ports (ie no firewall blocking). The client UI should at least support Mac and Windows. The server can be any of the three.

The above solution probably works, I havent tested it :). I am curious about alternatives. Will look into threads for which I seem to be duplicate. Thanks

First, some English. Estimation is not the same as measuring. If I wanted to measure, I would do so ;) (I might actually do that later). Also I have a dev question posed related to this post.

I would like to store data in a MySQL table, The data type of the table is date, long, long, long, long, short int , where i would have a composite key based on date, long, long.

There could be more indexes as well. (Edit) It seems that it would be 42 bytes of data plus the indexes. You can use 1 million records inserted every day for data growth. Should I just say 42 million bits would be added every day (I guess there is more, based on configuration)

So looking at this http://dev.mysql.com/doc/refman/5.1/en/numeric-types.html it comes out to be

I can have lets say 1 million rows added every day. How much would I rack up in Diskspace? Any references would be greatly appreciated (I will try to find some and post it here as well)

On a MySQL 5.1 Enterprise server db-dump takes a lot of time. I would like to reduce the amount of time to restore the db-dump or to be able to quickly go back to a known state. This machine is used for testing. So lets say we have a DB Dump at point A. I do some changes to the DB with an application and after a while I want to go back to the same db dump. I have the following options.

• Delete every thing and start with a fresh db-dump restore, takes about 14 hours
• Use a backup of the entire mysql folder

Do I have any other options. Some way to quickly get back to state A. Or to speed up my dbdump. I will try to provide you with more details if you ask (I am not sure what should I be putting here).

I am trying to boot CloneZilla on Xserver 2.1 from CD. I have tried to use C key as well as option and command while booting. It always goes to MacOS instead.

I have also tried to tinker with the second button present on the front panel with the following procedure as stated in the Apple Xserve manual

**Choosing a Startup Method from the Front Panel**
You can use the Xserve front panel controls to choose from alternative startup 
methods that might be helpful in special circumstances.
To choose a startup method from the front panel:
1 With the power off, hold in the system identifier button while you press the 
on/standby button.

2 Continue to hold in the system identifier button until the top row of blue lights   blinks sequentially.  3 Release the system identifier button, and then press it again repeatedly to turn on the       light that corresponds to your chosen startup method:

4 When you’ve made your choice, hold in the system identifier button until all lights in 
the top row are on, and then release.
The Xserve starts up using the chosen method

I have a PPTP based VPN setup on a remote machine. This machine is a Windows 2008 R2 Datacenter edition VM.

The VPN connection does not have DHCP enabled and does not have an IPv4 default gateway.

Now with that I would like to use only in the instance of a particular protocol (say Voip or Bittorent). So if traffic originates from my machine it can use its default gateway and public internet to go out and come back. But for some instances I would like this traffic to go/come_back via VPN so from an outside perspective this traffic is not from my system (assuming that eavesdropping on VPN connection is not feasible due to encryption)

I have downloaded mysql-advanced-5.1.46sp1-osx10.6-x86_64.dmg and installed it. I have a custom my.cnf file and no data to begin with. I dont see mysql_install_db any where. How can i install one. The server where I am copying this configuration from has mysql_install_db present. pkginfo returns nothing useful and man page states that this mysql_install_db is from Oracle.

I would like to make a script which can maintain multiple servers via SSH. I want to control the authentication/authorization in such a manner that authentication is done by gateway and any other access is routed through this ssh server to internal services without any further authentication/authorization requirements.

So if a user A can log into server_1 for example. He can then ssh to server_2 without any other authentication and do what ever he is allowed to do on server_2 (like shut down mysql, upgrade it and restart it. This could be done via some remote shell script).

The problem that I am trying to solve is to come up with a deployment script for a JavaEE system which involves databases and tomcat instances. They need to be shutdown and re-spawned. The requirement is to have a deployment script which has minimal human interaction as possible for both developers and operation.

Is it a good idea to run tomcat (tcServer from spring source) directly on port 80 (or 443) as root

One could use mod_proxy or mod_jk from apache2 to work around it. I have also seen iptables based solution (ie iptables -t nat -A PREROUTING -p tcp --dport 80 -j REDIRECT --to-ports 8080)

Are there any pros/cons which one should know when it comes to production use?