I'm running several virtual machines on VMware ESXi (mostly Centos but few are Ubuntu) and it seems like there are two different ways of dealing with memory allocation. Here are two examples:
(a) allocate 4GB memory to virtual machine and create 2GB swap file.
(b) allocate 6GB of memory to virtual machine and no swap file.
What are pros and cons of these approaches?
I'm running Vmware ESX 4.1 and I have a development VM that I recently upgraded from Ubuntu 10.04 to 11.04. Then I tried to re-install VMware Tools and some of the modules gave me an error and would not compile. As a result I'm having problems with backing up this virtual machine now and I suspect VMware tools is the reason. I installed latest patches for VMware host, that included an update to VMware Tools (v8.3.7 build-381511) but I'm still getting the same error.
The error I'm getting is like this:
...
/tmp/vmware-root/modules/vmhgfs-only/super.c:73:4: error: unknown field \u2018clear_inode\u2019 specified in initializer
make[2]: *** [/tmp/vmware-root/modules/vmhgfs-only/super.o] Error 1
make[1]: *** [_module_/tmp/vmware-root/modules/vmhgfs-only] Error 2
make[1]: Leaving directory `/usr/src/linux-headers-2.6.38-8-generic'
make: *** [vmhgfs.ko] Error 2
make: Leaving directory `/tmp/vmware-root/modules/vmhgfs-only'
and also this:
/tmp/vmware-root/modules/vmci-only/vmci_drv.c:91:4: error: unknown field \u2018ioctl\u2019 specified in initializer
/tmp/vmware-root/modules/vmci-only/vmci_drv.c:91:4: warning: initialization from incompatible pointer type
/tmp/vmware-root/modules/vmci-only/vmci_drv.c: In function \u2018vmci_init\u2019:
/tmp/vmware-root/modules/vmci-only/vmci_drv.c:151:4: error: implicit declaration of function \u2018init_MUTEX\u2019
make[2]: *** [/tmp/vmware-root/modules/vmci-only/vmci_drv.o] Error 1
make[1]: *** [_module_/tmp/vmware-root/modules/vmci-only] Error 2
make[1]: Leaving directory `/usr/src/linux-headers-2.6.38-8-generic'
make: *** [vmci.ko] Error 2
make: Leaving directory `/tmp/vmware-root/modules/vmci-only'
Any ideas?
I have an LTO-4 tape connected to a Linux host and Linux creates all the usual devices for this tape as follows:
/dev/nst0
/dev/nst0a
/dev/nst0l
/dev/nst0m
/dev/st0
/dev/st0a
/dev/st0l
/dev/st0m
I normally use /dev/st0 and /dev/nst0 devices for writing to the tape but now I want to try hardware compression. I assumed that I simply need to use one of the other devices but here is the question - which one? I read trough documentation and I even looked at source code of the driver but I could not find a straight answer.
I have a piece backup software that supports two options for backing up VMware virtual machine - one is just backing up VM's files (taking a snapshot first) and the other using vStorage API. The first option can backup VMs from the free version of ESX(i) the other only works with the paid version. So, I see an immediate disadvantage of using vStorage API - we would have to pay $500 for VMware license, but I assume there should be something good that compensates for this, right?
VMware is avoiding the fact that it is really easy to backup VM files if your storage is on NFS, so they keep quiet about this and only promote vStorage API. I have a feeling that there is nothing substantial in this API and they just want more money. Could someone speak from experience and shed some light on REAL advantages of using vStorage API for backups?
I want to remount a file system with optimized options (noatime,nobarrier,etc.) on SLES11. I was going to do it by updating /etc/fstab first and then do something like mount -o remount /srv/share. The problem is that there are images of several active virtual machines on this file system exported via NFS and I was thinking if remount can be done without shutting them down. So is remount operation disruptive? Have anyone ever tried something like this before? What was the outcome?
I'm building high capacity storage for video. It is currently at 20TB mark and keeps growing. Currently, I have 12-bay enclosure filled with 2TB drives and configured in RAID-6. I'm considering to buy another 12-bay enclosure and fill it with 2TB drives as well and I'm considering the following 2 options:
In first case I will be facing ~4-6 weeks of rebuild time before I can use extra capacity. Also, if I need to add another shelf to this array later and use same approach the rebuild time will likely be even longer.
In the second case I'm loosing some capacity and potentially performance but it seems easier to handle and availability times are more predictable.
I also plan to grow this array to 7-10 shelves in total within next 2-3 years and the entire array will comprise one huge volume formated with XFS. So, could anyone speak from their experience and elaborate on pros and cons of each approach and potential problems I will be facing as time goes by.
Thanks