marcelo_soares's Posts

Visitá vmsoares.wordpress.com para saber más acerca de mis preparativos para el VMworld 2012! Visite o blog vmsoares.wordpress.com para saber mais sobre os meus preparativos para o VMworld 2012!

Hola a todos! Solo para informarles que estoy haciendo un blog en español/portugues con mis insights y experiencias, arrancando con una cobertura del VMworld 2012 completa: vmsoares.wordpress.com Saludos!!!!!

Yesterday got this problem. There is a (very unclear) KB article on that: http://kb.vmware.com/kb/2001550 This problem resides on the fact that the ESXi Server uses several RAMDISKS, each one have its role on the COS. The "root" stores all the information under / that is not linked to another partitions. If you have this issue, perform the command "vdf -h" to check if the root disk is full or almost full. We (my co-worker Gabriel was originally with the problem) could see that the use of the root ramdisk was 99%, with only 16k left. Any operation seems to fail (create files, start services, etc). Navigating to the root partition, we could see that there was an /ibm directory, and the scratch link was pointing to it. This was an IBM embedded image of ESXi 5, and seems IBM is doing things wrong. You can NEVER point the scratch partition to the ramdisk itself - you only have 32MB, and this way this tiny space will be used to store long term logs, generate log bundles, store cores, etc. To resolve the proble, backup the files on /ibm partition (if you want to keep old logs) and erase them. Restart management services on the ESXi ("services.sh restart" will do the job) and you are ready to go. To permanently resolve the issue, redirect the scratch to an persistent storage: http://kb.vmware.com/kb/1033696 Cheers!

Just passed the VCP 5 and have some tips for you interested on having the exam: - Don't care about maximuns, there were not a single question about it. - Take a look on what the %WAIT, %READY and %USED CPU counters means, you need to understand what each of them meand on the performance tab of vCenter. - Resource pools shares are also important to understand. Effects on VMs on children pools, shares, reservations, etc. - There were a couple os questions about VSA administration/installation, and also about VDR. - Know about plugins, how to configure them, and how they appear on VI Client. - Check every information displayed on summary tab of hosts and VMs. Specific information about them may appear. - Know about MRU/FIXED policies, and how each behave. Check the manage paths window on how the paths appear for each policy. - Check what logs can be seen on the DCUI of an ESXi. - Upgrade from 3.x, 4.x to 5 appear also. Be comfortable with them. - Think and read each question at least twice before answering. Some are very tricky. Be sure to pass on the Mock Exam at mylearn.vmware.com, and also on the online exams of Simon Long: http://www.simonlong.co.uk/blog/vcp5-practice-exams/ (donate if you pass, the guy really deserves). Hope this helps Any questions, leave me a comment.

Entering on Google today, I could notice (happy to know what it was about) a funny doodle: a Turing machine. In my early times as Computer Science student, I needed to study "Computer Theory", which by the name would seem a complete waste of time, but where you can learn the basis of the things that are part of our actual IT jobs. As in any field of knowledge, learning the basis is to know better how things work. The knownledge of the latin language can help you understand portuguese or spanish better, knowing C will help you understand the Linux kernel, and if you already played around with a turing machine, you will understand general machines as a whole much better. Physical, virtual, cloud ones. I would recommend, to the ones that didn't seen much, any reading on Alan Turing's achievements and life. And also, learn about the incredible bomb he designed (with a great contribution of poland Marian Rejewski) to defeat the german Enigma machine, and the subsequent implementation of the Turing Machine idea. Well... at this precise moment, I am thinking on the last part of the second round of google's turing machine as you can see above: Try having some fun with that

If you already tried to read a VMware syslog generated file with Linux, you may have faced some difficulty due to the special characters <XXX> messing up the log file. I also wrote a post on how to remove it from a single file: http://communities.vmware.com/people/marcelo.soares/blog/2012/05/15/replacing-characters-by-newline-with-sed But... sometimes, you have around 20 files to analyze... so, here is an optimization: for i in `ls *.log| awk '{print $NF}'`; do cat $i | sed 's/<166>/\ /g' | sed 's/<181>/\ /g' > $i.ok; done ; rm -f *log This will have the work done in a few seconds

When doing a datacenter migration, you may need to register all VMs back to the vCenter/ESX inventory (or whatever reason you have to re-register VMs to an ESX). Here are the command lines to do so: Classic ESX: cd /vmfs/volumes; find ./ -name "*vmx" | sed 's/\.\//vmware-cmd -s register .\//g' | sh ESXi: cd /vmfs/volumes; find ./ -name "*vmx" | sed 's/\.\//vim-cmd solo\/registervm \/vmfs\/volumes\//g' | sh Change it whenever you want.

Do you need to test a Windows VM and starve its CPU? Simple! 1. Create a file named test.bat anywhere; 2. Type something like this inside it: @echo off :again echo Looping... goto again 3. Save it, go to cmd and to the "anywhere" directory. 3. Then run it typing test.bat <ENTER>. CPU spike of 100%....

Got some syslog files without CR/LF characters to analyze... only a "<166>" string on the place of the newline. If needed, you can use this simple script to resolve it: cat syslog.4.log | sed 's/<166>/\ /g' > syslog.4.log.ok (there is really an newline on the command, so the <166> will be replaced by a newline...

Just corrected my telnet.py script posted on http://communities.vmware.com/people/marcelo.soares/blog/2012/04/26/telnet-for-esx-and-esxi. Now with some improvements, typo correction and error detection Hope this helps anyone needing to troubleshoot network issues. Updated python script here: http://communities.vmware.com/servlet/JiveServlet/download/38-87947/telnet.py

Well, everyone already sat in front of an ESXi with a networking connectivity problem. Sometimes, ESX can't connect to vCenter (which is very well troubleshooted on the KB http://kb.vmware.com/kb/1029919) on 902 UDP, but sometime you want to check if ESX can reach a specific TCP port on another host (ESX or not). Example: to check if ESX can connect on an AD for authentication, NFS communication, host to host migration, etc. So I have created, based on the udp_client.py, a small telnet.py script that connects to a specific IP and port and tell if it is opened or not. Simple and easy. Usage: python telnet.py <IP> <port> Example of good connection: ~# python telnet.py 192.168.0.1 80 Host: 192.168.0.1 Port: 80 Sent 1 byte over port 80 to 192.168.0.1 Example of a bad connection: ~# python telnet.py 192.168.0.1 81 Host: 192.168.0.1 Port: 81 Not possible to connect to 81 on port 192.168.0.1 It needs some enhancements, but for sure will help on network troubleshooting!!! NOTE: Updated script here: http://communities.vmware.com/servlet/JiveServlet/download/38-87947/telnet.py Enjoy!!!

Well... who needs to troubleshoot ESX log files sometimes needs to uncompress several tgz files. So, here is a command line that may help on this: for i in `ls *.tgz| awk '{print $NF}'`; do tar -xvzf $i; done You don't need to specify any filename. Only with this all tgz files on a directory will be uncompressed.

Just finished a compilation after some time experiencing the conversion of paravirtualized XEN VM's to ESX 4.x. Having a lot of fun, obviously... You can find on the attached PDF the full article on doing the operation: http://communities.vmware.com/servlet/JiveServlet/download/38-62151/XEN_VMware_migration.pdf Any questions and insights are welcome.

Just playing around today with VDR and could find out that, when VDR VM have access to the VM VMDK's it is backing up, it mounts them on the VDR appliance instead of simply copying files over network (like VCB does with the nbd option). Is a way I never thought it could work. Just a problem happened when I tested to restart the appliance on the middle of a backup - the VM snapshot were maintained and the VMDK's remained mounted - little dangerous Dismounted the disks and removed the snapshots and everything came back to normal.

Saying you never was into a support session and suddenly need a disk driver floppy disk, or to copy an amount of data (GB's) from your computer to a VM without using direct network connection is a lie... it happens to all of us (at least the tech support slaves like me). Here some tools that may help on the day to day creation of images: Rawwrite: old and good image creation software for windows (linuxers please don't read all of this and go dd your data), is simple and I remember using it whan I was 15 and needed to install Linux with that old cdrom drive. Dowload it from here: http://www.chrysocome.net/rawwrite DoISO: burn and create ISO's, simple and effective, not too old as the last one, but useful as well: http://opensource.ebswift.com/DoISO/ For that ones who can't boot a Windows 2000/XP or other OS that does not comes to my mind now, use the VMware SCSI floppy: http://download3.vmware.com/software/vmscsi-1.2.0.4.flp Today it happened, almost needed a solution like this: had a customer who couldn't see VM hard disks using a Mcafee antivirus boot CD. Luckly using BusLogic disk adapter at the VM resolved the problem. LSI was not supported by the Mcafee OS. See you all PS: This week the most active member

You may alredy have experienced this: your ESX Server (mainly the 3.x series) becomes unresponsive, but all VMs are still up. You can't connect using VI client, nor manage it from vCenter, sometimes you get SSH and sometimes not. If you have ESXi, worst - it have no SSH connection (for default). Mainly, there are some reasons for that: - Your ESX is so overloaded with VMs that the COS can't have enough resources to process. This is sot so likely, but can happen (I've never seen one in front of mine, but is a possibility); - You have some management agent installed (from HP, Dell, IBM, etc) on the COS and it is, for some reason, consuming all COS CPU and/or memory; - CIM server are receiving a certain number of connections that makes it crash and use all COS memory; - Maybe other reasons... In the 3rd case, you will only be saved if you have SSH to host, or physical access. You can try disabling CIM by doing: ESX classic: - Type the following at the SSH session as root or physically at the host console logged in as root: service wsmand stop service pegasus stop service slpd stop ESXi: - Enter in unsupported mode (at physical console hit alt+f1 and type the word "unsupported" and type the root password) - Type the following: /etc/init.d/sfcbd-watchdog stop /etc/init.d/wsmand stop /etc/init.d/slpd stop This should make your ESX run better. After that, try restarting the management agents. If you want to know why it happens, or need further assistance, post a message con the communities forum and I am really sure someone will help you (maybe me...). Happy virtualization!