Skip to main content

RAM Disk Full Due To Inodes In ESXi Host




Last week I got into issue where one of my ESXi host was prompting error while creating virtual machine.

I verified the task and events of ESXi host and found below error generating while creating
Virtual machine "A general system error occured: Failed to open "/var/log/vmware/journel/ for write: There is no space". While further digging, I identified the ESXi host from where its generating this error from Task and Event section in vCenter.

Tried to take SSH of the question ESXi host but its was inaccessible via putty. However, ESXi host was up and running fine. Last option left to access the ESXi host from its management console which is ILO as its ESXi installed on HP server.

Also you can try accessing the SSH via other ESXi host or Linux machine using below command ssh -T servername. However, this will not give your prompt but you can type the command to get the output, but that was not giving any luck at that time so we use ILO for further troubleshooting.

Took the Shell access of ESXi host from ILO by pressing ALT+F1 to move into command prompt.
As initially it’s showing space issue, verified the disk space by using df command.

========================================================================
root# df /var/log/vmware/journal
Filesystem           1K-blocks   Used Available Use% Mounted on
/dev/mapper/db_vg-db  25663804 471036  23882460   2% /storage/db
========================================================================

From the above command it’s seems there is sufficient space on the disk partition where journal is writing the transactions. Then where the issue is...

I decided to verify the logs as its the gold mine to fine the issue. While going through vmkwarning.log I found multiple error like below
"Cannot create file /var/run/vmware/tickets... ...for process hostd-worker because the inode table of its ramdisk (root) is full"

So, from the logs it’s clear that the issue is not persisting from disk space issue, rather its issue with inode.

Now what is inode?

An inode is an entry in inode table, containing information ( the metadata ) about a regular file and directory. An inode is a data structure on a traditional Unix-style file system such as ext3 or ext4.

Node number also called index number, it consists following attributes.
File types ( executable, block special etc )
Permissions ( read, write etc )
UID ( Owner )
GID ( Group )
FileSize
Time stamps including last access, last modification and last inode number change.
File deletion time
Number of links ( soft/hard )
Location of ile on harddisk.
Some other metadata about file.

Now, we need to identify how many inodes are used by the file system and how many are available at present.

To identify this we can run the command    stat -f /

Also, you can run the command localcli system visorfs ramdisk list to identify the available Inode into the root partition.

outcome of the command

ID: 5        Namelen: 973     Type: visorfs
Block size: 4096
Blocks: Total: 449852     Free: 445356     Available: 445356
Inodes: Total: 8900       Free: 0

After running the command I found out of 8900 Inodes all 8900 Inodes are in use and 0 available. 

From this its crystal clear that its the issue with exhausting of Inodes.

Now, it’s time to identify which file is using high Inodes and how to clear them to make Inodes available.

Starting from /var| wc -l found its using around more than half of the Inodes. Now to dig deeper need to find which file under /var its using high Inodes. Under the /var/run i identify that the sfcb "/var/run/sfcb" is the file which is using high nodes.

sfcb is basically the hardware monitoring services which seems not to be critical but important to know the hardware status.

To make free the inodes from sfcb file I use VMware KB

Run this command to stop the sfcbd service:
/etc/init.d/sfcbd-watchdog stop

Manually delete the files in the var/run/sfcb directory to free inodes.

Run these commands to remove the files:

cd /var/run/sfcb
rm [0-2]*
rm [3-6]*
rm [7-9]*
rm [a-c]*
rm [d-f]*

Run this command to restart the sfcbd service:
/etc/init.d/sfcbd-watchdog start

Now restart the ESXi host management agents.

After deleting the sub files from sfcb folder I again verify "/var/run/sfcb | wc -last" and now its have only bunch of inodes connected to it.

Now, we retry the VM creation and it’s creating without any error.

Thanks for reading the post .. Happy sharing... :)

Comments

Post a Comment

Popular posts from this blog

Changing the FQDN of the vCenter appliance (VCSA)

This article states how to change the system name or the FQDN of the vCenter appliance 6.x You may not find any way to change the FQDN from the vCenter GUI either from VAMI page of from webclient as the option to change the hostname always be greyed out. Now the option left is from the command line of VCSA appliance. Below steps will make it possible to change the FQDN of the VCSA from the command line. Access the VCSA from console or from Putty session. Login with root permission Use above command in the command prompt of VCSA : /opt/vmware/share/vami/vami_config_net Opt for option 3 (Hostname) Change the hostname to new name Reboot the VCSA appliance.   After reboot you will be successfully manage to change the FQDN of the VCSA . Note: Above step is unsupported by VMware and may impact your SSL certificate and face problem while logging to vSphere Web Client. If you are using self-signed certificate, you can regenerate the certificate with the
2 comments
Read more

Issue : Configure Management Network option is Grayed out into ESXi

Last week I got into an issue of one of my client into Vsphere environment where one of its ESXi went done out of the network. Issue was IP address was showing 0.0.0.0 on main Esxi screen and when I tried to change the network configuration, its " Configure Management network option was greyed out.  I tried to gid into it and try to analyis its vmKernal and vmwarning logs. What I found is its VMkernal switch got removed due to unexpected reason. So to resolve the issue I tried to reconfigure its vswitch0 (vmk0) by going into Tech Mode of that Exi. Below are the steps which I followed to resolve the issue. 1) Login to ESXi 2) Press F2, Check if you " Configure Management network " is greyed out or not" if yes,    follow below 3) Press ALT+F1 to move the ESXi screen to tech mode   ( This is command line like dos) 4) login with root account 5) Run the following command into it esxcli network ip interface add --interface-name= vmk0
2 comments
Read more

Collecting Logs from NSX-T Edge nodes using CLI

  This article explains how to extract the logs from NSX-T Edge nodes from CLI. Let's view the steps involved: 1) Login to NSX-T  Edge node using CLI from admin credentials. 2) Use of  " get support-bundle " for Log extraction. get support-bundle command will extract the complete logs from NSX-T manager/Edge nodes. nsx-manager-1> get support-bundle file support-bundle.tgz 3) Last step is to us e of " copy file support-bundle.tgz url " command. copy file will forward your collected logs from the NSX-T manager to the destination(URL) host from where you can download the logs. copy file support.bundle.tgz url scp://root@192.168.11.15/tmp Here, the URL specified is the ESXi host ( 192.168.11.15) under /tmp partition where logs will be copied and from there one can extract it for further log review. Happy Learning.  :)
Post a Comment
Read more