Base VM is a Debian Jessie, 3.16 kernel with local ssd mounted for mysql (/var/lib/mysql).
Few weeks ago I created a snapshot from this base VM and wanted to create a new VM with more power (more CPU and memory) and then I ran into this problem (reported by me): https://code.google.com/p/google-compute-engine/issues/detail?id=202&colspec=ID%20Type%20Component%20Resource%20Service%20Status%20Stars%20Summary%20Log
As there was no response or resolution on this ticket we dropped this upgrading but now our base VM crashed and seems like to be similar issue. Trace: http://pastebin.com/sq4EMNPb
Is it some known bug? Is it some settings causing this? Or should we drop using local SSD and just continue with peristant ssd?
All ideas and opinions are welcome!
This is a known issue affecting only the NVMe driver. Google is working on it, and recommends you use Local SSD with the SCSI interface as a temporary workaround if you are running into this.