Returning to the Spice for Mac problem. Also working on a memory allocation patch. Updates on the Linux hosts and VMs backups.
This morning, my Synology was completely unresponsive. Investigating a bit, I noticed two things:
- The Snapshot Replication tasks were piling up. I think a VM file is changing too quickly for snapshotting to work OK.
- On both Big and Muse, there were dozens of
rsyncjobs piling up… Ooops 😦
So I decided to change my backup strategy:
- I modified my backup scripts to use
flock, so that only one
rsyncper machine is active at a time. Otherwise, since I’m using
rsync --deleteto make sure that my backup does not keep files I removed locally (keeping them in the history is why I have snapshots), two simultaneous
rsynccollide into one another, one deleting the temporary files that the other creates. Ouch. Using
flockallows me to keep a high backup frequency (every 5 minutes), while ensuring that if a large file is modified, enough time is spent copying it.
- I moved the backups of VM image files to a separate shared folder, with a much lower snapshot replication schedule (once a day instead of every 5 minutes).
Later notes were lost because I saved from a window that had an older version of the post 😦 From memory, there was:
Investigation of a problem regarding tearing with video rendering, which Marc-André says is a problem on the Mesa side.
Investigation of networking issues with my guests. Apparently, when “something” happens outside, e.g. DNS / DHCP server not responding or something like that, then I lose user bridged networking, and have to reboot. Investigation later showed that the tap0 and tap1 interfaces are gone. Not sure why. Rebooting fixes it, but I wish there was a better way.