Error "The file table of the ramdisk 'var' is full. As a result, the file /var/###### could not be created by the application '######'."
Article ID: 376243
Updated On:
Products
Issue/Introduction
- Tasks fail with an error
The file table of the ramdisk 'var' is full. As a result, the file /var/###### could not be created by the application '######'.
- vm-support command fails to generate a log bundle.
Traceback (most recent call last): File "/bin/vm-support", line 21, in <module> logging.getLogger('').addHandler(getLogFileHandler(VMSLOGFILE, logging.INFO)) File "/lib64/python3.8/site-packages/supportbundle/cli/utils.py", line 16, in getLogFileHandler File "/lib64/python3.8/logging/handlers.py", line 148, in __init__ File "/lib64/python3.8/logging/handlers.py", line 55, in __init__ File "/lib64/python3.8/logging/__init__.py", line 1143, in __init__ File "/lib64/python3.8/logging/__init__.py", line 1172, in _openOSError: [Errno 28] No space left on device: '/var/log/vmsupport.log'
- You see many .trp files under /var/spool/snmp direcotory even though SNMP is not enabled.
vdu -acommand shows /var/spool/snmp directory consumes a large number of inodes.
For '/var/spool/snmp': ramdisk var: 30547968 ( 7459 inodes)
Cause
/var/spool/snmp directory is unexpectedly created on starting hostd and filled up with many .trp files. As snmpd is disabled, it is unable to process .trp files and because of that host is running out of i-nodes on /var partition.
Resolution
This issue is fixed in ESXi 7.0 Update 3o and 8.0 Update 1c.
As a workaround, remove /var/spool/snmp directory.
# rm -rf /var/spool/snmp
Additional Information
You see trap files in a SNMP directory under /var/spool even though SNMP is not enabled
After the hostd service starts, for example after an ESXi host reboot, it might create a SNMP directory under /var/spool and you see many .trp files to pile up in this directory.
This issue is resolved in this release. The fix makes sure that the directory /var/spool/snmp exists only when SNMP is enabled.
Feedback
