Hi,
We've got a cluster of ESXi 5 hosts (build 623860) and one of them is generating a bunch of messages like:
'The ramdisk 'root' is full. As a result, the file /var/log/ipmi/0/.sel_header.raw could not be written'.
The actual message differs from message to message but all of them start with the fact that the ramdisk is full. I've been googling a bit and investigating a bit and it turns out that the ramdisk is 99% full. I wasn't sure why this host was different but after comparing things, it looks like there's a config difference with respect to log files. The other hosts appear to have their log files (/var/log) redirected to the /scratch directory.
The questions I have are:
- why would one host be setup differently to the others for locations of log files (they're all the same version build)? What I mean here is why are the other host log files redirected to /scratch when this one is not. Could it just be a case of using a VMWare build vs using a customised vendor build? The person who set these up is not available to ask and I doubt they manually altered the default build settings to redirect logs.
- can this host config be changed easily to the same as the other hosts (could possibly change the log files to symbolic links point to scratch dir but is there a better way to do it)?
- if I want to just free some space and move the archived log files, which ones are safe to move? I read somewhere that anything ending .gz is 'archived' and can be moved safely (certainly I think hostd and vpxa gz files look like they can be moved but there are other like boot.gz that I'm unsure about)
These hosts will most likely be replaced with new hardware in the next couple of months so a work-around is fine as a solution. I might just move the archived logs to /scratch as a work-around.
Cheers,
Tim
