In order to find the reason of some seconds of data loss of last wednesday ( https://logbook.virgo-gw.eu/virgo/?r=58547 ) on olserver52 DAQ framebuilder we performed the same actions on the VMware Fault Tolerance (FT) infrastructure that we did last tuesday and wednesday.
Filesystem affected could be in principle: /olusers, /virgoData, /virgoLog, /virgoApp, /virgoDev.
Chronology - baseline status is with FT paused
9:25 LT: start of Fault Tolerance resuming (a fs01 disk image replica is built transparently)
... no appreciable effects spotted on nfs clients performance
9:57 LT: Fault Tolerance active (the disk image replica has completed and a standby replica of the fs01 VM is synchronized in lockstep in realtime) ; a glitch occurs on the olserver52 network output flux
... load on fs01 increases 5 times, olserver53 cpu IOwait increases from 0 to 1.25% , there are no effects on the stolxx writing servers
10:45 LT: Fault Tolerance is paused and left in this state (the fs01 standby VM and disk replica is destroyed); a second glitch appears on the olserver52 network output flux