Server Disk Statistics and VPS's

[Xao]

I recently moved our mail server (and upgraded to 5.0) into a OpenVZ VPS.

Since I did that, the disk statistics are no longer working for Disk Usage under the Admin interface.

I think this may be related to '/' not being mount on anything like /dev/md0 but it is mounted on a 'simfs' I was wondering if anyone else was experiencing this and had a work around?

I added a few entries into my 'fstab' as follows:
simfs / simfs rw,noatime 0 0
proc /proc proc rw,nodiratime 0 0
none /dev/pts devpts rw 0 0
devpts /dev/pts devpts rw 0 0


My /etc/mtab reads:
simfs / simfs rw,noatime 0 0
proc /proc proc rw,nodiratime 0 0
devpts /dev/pts devpts rw 0 0
proc /var/named/chroot/proc proc rw,nodiratime 0 0

[brian]

/opt/zimbra/libexec/zmdisklog is the script that is responsible for checking disk space. Currently it only reports on ext3 filesystems.

You can hack it to change the
mount -t ext3 command to be something like mount -t simfs

[Xao]

Thanks, I will do that, do I have to do anything else to clear out any settings for the old graphs?

The old graphs still show (historical) for /dev/md0 and /dev/md2 which no longer exist, or will it automatically build new ones?

[brian]

It will build new ones, and I believe it will clear the older ones after some period but I don't recall what this is right now.

[Xao]

Well, I modified the script from ext3 to simfs, but the script comes back with:

[root@webmail signup]# /opt/zimbra/libexec/zmdisklog
df: `simfs': No such file or directory
[root@webmail signup]#

Anyone else have a VPS working with diskstats? Or heck, even iostat doesn't appear to work in a VPS.

[ArcaneMagus]

As far as I can tell what the script does is run "mount -t" and then whatever types you input after that. This will output something like the following:

/dev/sda2 on / type reiserfs (rw,acl,user_xattr)
/dev/sda1 on /boot type ext3 (rw,acl,user_xattr)
/dev/mapper/system-opt on /opt type reiserfs (rw,acl,user_xattr)

It then feeds this to a processor that will separate out the device name and the mount point for each line that the mount command output. I will use "/dev/mapper/system-opt on /opt type reiserfs (rw,acl,user_xattr)" in the following examples.

After separating out the device name and the mount point it then builds the following line:

df -mlP device name | tail -1

so in my case the command it will build is:

df -mlP /dev/mapper/system-opt | tail -1

Which outputs a line like:

/dev/mapper/system-opt       102397     16346     86052      16% /opt

This line is then parsed to grab the total blocks, used blocks, and available blocks
After this it does some housekeeping on all the variables and finally writes a line like:

info 2008-04-03 22:09:29 DISK: mail: dev: /dev/mapper/system-opt, mp: /opt, tot: 102397 avail: 86052

to the log file.

My guess is that the df -mlP command is not liking something that you have done...try running through this manually to see if you can see the error.