Hi all,
Last night one of my ESX hosts dropped out of VirtualCenter. All the VM's stayed running on it and I could get onto it. After a reboot it lost all networking info so I rebuilt that from the iLO...
What I have noticed is this...
Filesystem Size Used Avail Use% Mounted on
/dev/cciss/c0d0p2 4.9G 4.9G 0 100% /
Not sure when this filled up and by what, a little unsure how to find out? ls just lists about 128K in there!
Any ideas?
UPDATE: Found the files in /var/core! Lots of them! Why would I have so many core dumps? My server has not PSOD'ed at all?!?!
I'm afraid not... Each time hostd crashes, a dump is created for support purposes..
Just watch the directory when something suspect is happning
/Rubeck
Hi..
Check to see if /var/core is full of hostd dumps... If yes.. delete them, and watch the server for a while to see if this is ongoing due to hostd problems.
/Rubeck
Thanks Rubeck,
There was 2Gig's worth of core files and a fair number of them. When are these generated as there are a few!
There are no more now since deleting them all and that is a few hours after. Does ESX not manage these files and clear them out?
I'm afraid not... Each time hostd crashes, a dump is created for support purposes..
Just watch the directory when something suspect is happning
/Rubeck
A lot /var/core dumps are definitely showing a vmkernel or hostd issue. You may want to send one to vmware support to have them take a look at what is the cause.
I would also recommend you rebuild your hosts whenever possible as it looks like you used the default installation partitioning scheme which you've found out is not the best idea since /var is on the same partition as / and as soon as that fills, you can pretty much kiss your host goodbye.
I typically do something like this: (I may have missed some because I am going by memory)
/ - 5G - primary
/boot - 512MB - primary
/swap - 1600 - primary
/var 5GB -extended partiton
/vmcore - 110 - extended partition
/tmp -1024 - extended partition
I've also recently started utilizing the local disk for a vmfs partition (created through VC) to store the vm swap memory as I don't really need that on the SAN and I don't find it impacts VMootion performance that much (vs keeping it on the SAN).
Hello,
I do the following to avoid these issues:
/boot -> 200MB
/ -> 5GB
/var -> 5GB
/var/log -> 5GB
/tmp -> 5GB
/home -> 5GB
swap -> 2GB
The problem with having /var a part of / is that you will run out of space rapidly often before you delete things and often you do not want to delete those files until you take a look at them. If it was me, I would consider a different filesystem layout to alleviate the need to constantly delete possibly needed data, and to account for future updates.
Best regards,
Edward L. Haletky
VMware Communities User Moderator
====
Author of the book 'VMWare ESX Server in the Enterprise: Planning and Securing Virtualization Servers', Copyright 2008 Pearson Education.
CIO Virtualization Blog: http://www.cio.com/blog/index/topic/168354
As well as the Virtualization Wiki at http://www.astroarch.com/wiki/index.php/Virtualization
my 2 english pennies (over 2000+ ESX host deployments and 0 failures)
/boot -> 200MB
/ -> 8GB
/var -> 4GB
/var/log -> 5GB NOT ASSIGNED/CREATED
/tmp -> 4GB
/opt -> 4GB
/home -> 5GB NOT ASSIGNED/CREATED (script restricts large user usage)
swap -> 1600MB
/vmkcore -> 110MB
The Rest is VMFS3 local Store