[OmniOS-discuss] Got a larval fix for you, Bob!
Dan McDonald
danmcd at omniti.com
Tue Apr 4 05:37:31 UTC 2017
Are you bringing up and down zones this quickly in your deployment?
I appreciate the test case, and it appears that after 1:40, a second shutdown works:
bloody(~)[0]% /bin/time sudo zoneadm -z lipkg0 boot ; /bin/time sudo zoneadm -z lipkg0 shutdown ; /bin/time sudo zoneadm -z lipkg0 shutdown
real 0.9
user 0.0
sys 0.0
zone 'lipkg0': unable to shutdown zone
real 1:40.1
user 0.0
sys 0.0
real 6.0
user 0.0
sys 0.0
bloody(~)[0]% zoneadm list -cv
ID NAME STATUS PATH BRAND IP
0 global running / ipkg shared
- lx0 installed /zones/lx0 lx excl
- lx1 installed /zones/lx1 lx excl
- lipkg0 installed /zones/lipkg0 lipkg excl
- lx2 installed /zones/lx2 lx excl
bloody(~)[0]%
Alternatively, a one-second sleep appears to be enough to miss the race:
bloody(~)[0]% /bin/time sudo zoneadm -z lipkg0 boot ; sleep 1 ; /bin/time sudo zoneadm -z lipkg0 shutdown
real 1.1
user 0.0
sys 0.0
real 7.0
user 0.0
sys 0.0
bloody(~)[0]% zoneadm list -cv
ID NAME STATUS PATH BRAND IP
0 global running / ipkg shared
- lx0 installed /zones/lx0 lx excl
- lx1 installed /zones/lx1 lx excl
- lipkg0 installed /zones/lipkg0 lipkg excl
- lx2 installed /zones/lx2 lx excl
bloody(~)[0]%
I do believe that pwaiting on forked processes /usr/sbin/shutdown launches will eliminate more times this can happen, however. I will be putting such a fix out for review tomorrow on the illumos list.
Thank you for your help and sorry for the delay in addressing this,
Dan
More information about the OmniOS-discuss
mailing list