[DRBD-user] Select the cache configuration of KVM/DRBD for avoid lost of data

Cesar Peschiera brain at click.com.py
Mon Apr 14 17:44:30 CEST 2014

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.

Hi to all

If somebody can help me, will be very appreciated

I like to do a questions about of security of data and as get the best
preformance without risk of lost of data when HA is applied

In Proxmox with KVM i have this options of cache for my VMs:
1- No cache
2- Direct sync
3- Write through
4- Write back
5- Write back (unsafe)

My physical machines have:

- A single SATA HDD for use exclusively with DRBD (without a RAID
- LVM is on top of DRBD
- Bonding-rr with two NICs (that is working perfectly, NIC to NIC)
- A VM have his virtual HDD in the DRBD volume

The DRBD configuration:
- These are my manual configurations (summarized):
net: sndbuf-size 0; no-tcp-cork; unplug-watermark 16; max-buffers 8000; 
max-epoch-size 8000; verify-alg sha1;
disk: on-io-error detach; al-extents 3389; resync-rate 75M;

I want to know what i should configure in KVM and/or DRBD for that this VM
don't use the RAM as write cache, and it is considered that the data are
commited to disk when the data really is in the buffer of DRBD?.

A note: I have actually configured "cache=none" in my VM and the test of
volume replicated DRBD is doing once a week successfully, but i think that
with the option "cache=none" in my VM not guarantees that the data are 
always commited in the buffer of volume DRBD.

Best regards

More information about the drbd-user mailing list