WilliamLam.com

How to Persist Configuration Changes in ESXi 4.x/5.x Part 1

by // 12 Comments

The question of persisting configuration changes in ESXi and the expected behavior using the ESXi Shell is a question that comes up quite frequently. I think this is primarily due to user's experience with the classic ESX Service Console and the expectation that ESXi will behave exactly the same.

First off, any changes made to ESXi 4.x or 5.x using the vSphere Client, any of the vSphere SDK toolkits (vCLI, PowerCLI, etc) or going through vCenter will preserve the changes across system reboots. The caveat is when going directly into the ESXi Shell (formally TSM) and manually editing or adding files to the default ESXi installation, where the persistence of the files will vary depending on what was updated.

So why does editing files in ESXi filesystem not persisted across system reboots? The reason for this is ESXi uses an in-memory filesystem. For more details, check out The Architecture of VMware ESXi whitepaper by VMware. As I mentioned earlier, the persistence of files will depend on what was updated as there are exceptions to the rules.

There are certain configuration files (e.g. /etc/inetd.conf) that are automatically backed up through a cronjob which looks for particular files under /etc that have been marked with the stickybit. A user can not manually mark a file with the stickybit and have it automatically backed up, it requires one additional file which is implemented by the VisorFS. ESXi creates a copy of these stickybit files and renames the original as .#filename. The backup process will then look for any .#* files and back those up. Due to this special permission mechanism, you can not manually create/touch files with this format as explained by a VMware employee on this VMTN thread.

UPDATE (08/14/2014) - Take a look at this article for more details regarding the various times an automatic backup is performed by ESXi

The "automatic" backup process occurs in the following scenarios:

1) /sbin/auto-backup.sh runs via cron the first minute of every hour and performs a "find" operation under /etc looking for files that being with ".#":

~ # cat /var/spool/cron/crontabs/root
#syntax : minute hour day month dayofweek command
01 01 * * * /sbin/tmpwatch.sh
01 * * * * /sbin/auto-backup.sh #first minute of every hour (run every hour)

Here is an example of the files backed up in ESXi 4.x:

~ # find etc -follow -type f -name ".#*"
etc/dropbear/.#dropbear_dss_host_key
etc/dropbear/.#dropbear_rsa_host_key
etc/security/.#access.conf
etc/vmware/hostd/.#hostsvc.xml
etc/vmware/hostd/.#pools.xml
etc/vmware/hostd/.#vmAutoStart.xml
etc/vmware/ssl/.#rui.crt
etc/vmware/ssl/.#rui.key
etc/vmware/.#dvsdata.db
etc/vmware/.#esx.conf
etc/vmware/.#license.cfg
etc/vmware/.#locker.conf
etc/vmware/.#snmp.xml
etc/.#hosts
etc/.#inetd.conf
etc/.#rc.local
etc/.#chkconfig.db
etc/.#random-seed
etc/.#resolv.conf
etc/.#syslog.conf
etc/.#shadow
etc/sfcb/repository/root/interop/.#cim_indicationfilter.idx
etc/sfcb/repository/root/interop/.#cim_indicationhandlercimxml.idx
etc/sfcb/repository/root/interop/.#cim_indicationsubscription.idx
etc/sfcb/repository/root/interop/.#cim_listenerdestinationcimxml.idx

Here is an example of the files backed up in ESXi 5.x:

~ # find etc -follow -type f -name ".#*"
etc/vmware/vpxa/.#vpxa.cfg
etc/vmware/hostd/.#hostsvc.xml
etc/vmware/hostd/.#pools.xml
etc/vmware/hostd/.#vmAutoStart.xml
etc/vmware/ssl/.#rui.key
etc/vmware/ssl/.#rui.crt
etc/vmware/.#config
etc/vmware/.#esx.conf
etc/vmware/.#license.cfg
etc/vmware/.#locker.conf
etc/vmware/.#snmp.xml
etc/vmware/.#dvsdata.db
etc/sfcb/repository/root/interop/.#cim_indicationsubscription.idx
etc/sfcb/repository/root/interop/.#cim_listenerdestinationcimxml.idx
etc/sfcb/repository/root/interop/.#cim_indicationhandlercimxml.idx
etc/sfcb/repository/root/interop/.#cim_indicationfilter.idx
etc/sfcb/repository/root/interop/.#sfcb_registeredprofile.idx
etc/sfcb/repository/root/interop/.#sfcb_registeredprofile
etc/sfcb/repository/root/config/.#omc_namespaceconfig.idx
etc/sfcb/repository/root/config/.#omc_namespaceconfig
etc/sfcb/repository/root/config/.#omc_config.idx
etc/sfcb/repository/root/config/.#omc_config
etc/sfcb/.#uuid
etc/sfcb/.#sfcb.cfg
etc/ssh/.#ssh_host_rsa_key
etc/ssh/.#ssh_host_rsa_key.pub
etc/ssh/keys-root/.#authorized_keys
etc/ssh/.#ssh_host_dsa_key
etc/ssh/.#ssh_host_dsa_key.pub
etc/vmsyslog.conf.d/.#vmkernel.conf
etc/vmsyslog.conf.d/.#hostd.conf
etc/vmsyslog.conf.d/.#fdm.conf
etc/vmsyslog.conf.d/.#vpxa.conf
etc/security/.#access.conf
etc/.#chkconfig.db
etc/.#hosts
etc/.#ntp.conf
etc/.#random-seed
etc/.#resolv.conf
etc/.#vmsyslog.conf
etc/.#shadow

2) There is also another "internal" backup process that runs every 10 minutes which is not configurable that backups the ESXi filesystem. In the worse case, you may lose changes made in the last 10 minutes if you had a system crash and it reboots.

With the latest release of ESXi 5, ESXi Shell (previous known as unsupported mode and Tech Support Mode) is officially supported for troubleshooting purposes, modifying or adding files is still not officially supported. Though if you need to make changes or add custom files to either ESXi 4.x or 5.x, there are a few options to persist the changes.

Option #1 - If you edit/modify any of the files listed above, these are automatically backed up by ESXi, it's generally recommend you manually run /sbin/auto-backup.sh to ensure they are backed up after your changes.

Option #2 - Edit/Modify files using /etc/rc.local file which executes at the very end of the bootup process. You can make your file changes in this script and they will automatically execute once the system has ran through all the default start up scripts. There two different methods depending on the complexity of your modifications, you may choose to use something like a here-document to create new/modified files or copy configuration files which have been persisted through a local or remote VMFS/NFS datastore.

Here is an example of setting up SSH keys for ESXi and manually creating the .ssh directory and SSH public keys:

Here is an example of copying a script (ghettoVCB.sh) from persistent local datastore to /bin directory:

Once you have made your changes, you will need to manually run /sbin/auto-backup.sh which ensures the changes made to /etc/rc.local will be persisted upon the next system reboot.

Option #2 provides for the most flexibility and allows you to make various configuration changes or addition of files and can easily be integrated into kickstart installation. The entries would be part of your %firstboot stanza to generate the entries in /etc/rc.local file.

In part 2 of this article, I will go over another method that can be leveraged in both ESXi 4.x and 5.x where custom files can be added and persisted as part of the default installation process using kickstart or through a one time manual configuration. I will provide some examples including persisting custom firewall rules in ESXi 5, so stay tuned!

Hidden vCenter Debugging Performance Metrics

by // 1 Comment

While extracting the new performance metrics in vSphere 5 for a blog post, I came across a metric type that I had never noticed before, vcDebugInfo. These performance metrics seems to deal with some of the internal performance/counters in vCenter such as lock statistics, MoRef (Managed Object Reference) counts, etc. Majority of these metrics are available in either collection level 1 or 4 which is the highest level containing all statistics. VMware's best practice is to only enable collection level 1 or 2, 3 and 4 should only be enabled under VMware supervision for debugging purposes which is most likely when some of these stats may come in handy. So be warn, these are probably not supported by VMware

I found it interesting that these metrics are hidden from the vSphere Client UI, but they can easily be extracted when going through the vSphere API. It is just amazing on all the goodies you can find when going through the APIs 🙂

Metric Stat Level Description
vcDebugInfo
maximum.millisecond.activationlatencystats 4 The latency of an activation operation in vCenter
minimum.millisecond.activationlatencystats 4 The latency of an activation operation in vCenter
summation.millisecond.activationlatencystats 1 The latency of an activation operation in vCenter
maximum.number.activationstats 4 Activation operations in vCenter
minimum.number.activationstats 4 Activation operations in vCenter
summation.number.activationstats 1 Activation operations in vCenter
maximum.millisecond.hostsynclatencystats 4 The latency of a host sync operation in vCenter
minimum.millisecond.hostsynclatencystats 4 The latency of a host sync operation in vCenter
summation.millisecond.hostsynclatencystats 1 The latency of a host sync operation in vCenter
maximum.number.hostsyncstats 4 The number of host sync operations in vCenter
minimum.number.hostsyncstats 4 The number of host sync operations in vCenter
summation.number.hostsyncstats 1 The number of host sync operations in vCenter
maximum.number.inventorystats 4 vCenter inventory statistics
minimum.number.inventorystats 4 vCenter inventory statistics
summation.number.inventorystats 1 vCenter inventory statistics
maximum.number.lockstats 4 vCenter locking statistics
minimum.number.lockstats 4 vCenter locking statistics
summation.number.lockstats 1 vCenter locking statistics
maximum.number.lrostats 4 vCenter LRO statistics
minimum.number.lrostats 4 vCenter LRO statistics
summation.number.lrostats 1 vCenter LRO statistics
maximum.number.miscstats 4 Miscellaneous statistics
minimum.number.miscstats 4 Miscellaneous statistics
summation.number.miscstats 1 Miscellaneous statistics
maximum.number.morefregstats 4 Managed object reference counts in vCenter
minimum.number.morefregstats 4 Managed object reference counts in vCenter
summation.number.morefregstats 1 Managed object reference counts in vCenter
maximum.number.scoreboard 4 Object counts in vCenter
minimum.number.scoreboard 4 Object counts in vCenter
summation.number.scoreboard 3 Object counts in vCenter
maximum.number.sessionstats 4 The statistics of client sessions connected to vCenter
minimum.number.sessionstats 4 The statistics of client sessions connected to vCenter
summation.number.sessionstats 1 The statistics of client sessions connected to vCenter
maximum.number.systemstats 4 The statistics of vCenter as a running system such as thread statistics and heap statistics
minimum.number.systemstats 4 The statistics of vCenter as a running system such as thread statistics and heap statistics
summation.number.systemstats 1 The statistics of vCenter as a running system such as thread statistics and heap statistics
maximum.number.vcservicestats 4 vCenter service statistics such as events, alarms, and tasks
minimum.number.vcservicestats 4 vCenter service statistics such as events, alarms, and tasks
summation.number.vcservicestats 1 vCenter service statistics such as events, alarms, and tasks

How to Disable a vmnic in ESX(i)

by // 8 Comments

There was another interesting thread today on the VMTN community forums about disabling an unused vmnic from an ESXi host due to false alarms being generated from HP SIM. Due to the specific hardware version, disabling the vmnic through the BIOS was not an option and there were no alternatives from the discussion.

I decided to dig a bit to see what I could find and stumbled upon a neat little utility called vmkchdev (VMkernel Change Device?) and from what I can tell provides a method passing a particular device to be controlled by either the VMkernel or as a passthrough device to a virtual machine (think VMDirect Path).

Disclaimer: Please note, this is using an undocumented utility. You should test this out in a development/lab environment before using and you may want to also contact VMware support to get their blessings***

~ # vmkchdev
Usage:
-s (scan device)
-v (give device to vmkernel)
-p (give device to passthru/VM)
-l (list device state)
-L (list device state with details)
[0x][seg:[bus[:slot[.func]]]]

What I found while testing the utility is by passing it over as a passthrough device, the vmnic is actually unrepresented to the VMkernel and does not show up under network adapters or even the unused/unlinked adapter list in the vSwitch configurations. It seems that it is just masks the device away from the VMkernel as you can still see the active configuration in esx.conf and you can see the device listed using vmkchdev.

Here is an example of an unused physical nic vmnic1 that we would like to disable and unpresent to an ESXi host.

Here is the output from esxcfg-nics:

First we need to identify the vmnic's PCI slot, we do by running the "-l" or list operation and searching for the particular vmnice device.

~ # vmkchdev -l | grep vmnic1
000:002:01.0 8086:100f 15ad:0750 vmkernel vmnic1

Next we will pass the device from the VMkernel to passthrough/VM using the "-p" flag and specifying the PCI slot, which in this case it is 000:002:01.0. We will also need to refresh the network section so the changes are reflected in the vSphere Client by using vim-cmd.

~ # vmkchdev -p 000:002:01.0
~ # vim-cmd hostsvc/net/refresh

If we list our vmnic again using vmkchdev, you will notice the device is now owned by passthru versus the VMkernel.

~ # vmkchdev -l | grep vmnic1
000:002:01.0 8086:100f 15ad:0750 passthru vmnic1

Now if we check the output of esxcfg-nics and the vSphere Client, you will notice that vmnic1 is no where to be found

If you would like to enable or re-present the disabled vmnic, you just need to pass the device back over to VMkernel by using the "-v" flag.

~ # vmkchdev -v 000:002:01.0
~ # vim-cmd hostsvc/net/refresh

Your vmnic should now re-appear on all your screens and any existing NIC teams that may have exists is automatically restored. This trick actually works on both a used and unused vmnic

If you are trying to do this on ESX, vmkchdev actually has an additional option called "console" for the Service Console.

[[email protected] ~]# vmkchdev
Usage:
-s (scan device)
-c (give device to console)
-v (give device to vmkernel)
-p (give device to passthru/VM)
-l (list device state)
-L (list device state with details)
[0x][seg:[bus[:slot[.func]]]]

I found that you need to pass the vmnic from VMkernel to Console, passing it to passthru/VM will not work and an error is thrown if you do. Again, you can easily re-enable by passing it back to the VMkernel

vmkchdev -c 000:002:01.0

If you would like to automatically persist this change across reboots, specifically for ESXi as changes are not saved. You will need to add the following lines to /etc/rc.local which will execute the disabling of the vmnic's after bootup.

/sbin/vmkchdev -p 000:002:01.0
vim-cmd hostsvc/net/refresh

You will also need to run /sbin/auto-backup.sh to ensure the changes to /etc/rc.local are saved and reloaded upon the next reboot. For ESX, you can place it in /etc/rc.local without having to do anything extra as the changes persists across reboots for classic ESX

Now you can play hide and seek with your vmnic's without resorting to a system reboot or touching the BIOS. Though ideally, if you do have unused devices, you should definitely disable them in the BIOS if you have the option. On a side note, this might be a fun trick to play on one of your co-workers by hiding all vmnics 😉