December 2010 Archives

rebooting hamper to add new disks

| | Comments (0)
we're using the 'xm save' capabilities of xen, so other than 10-20 mins. of downtime, you shouldn't detect a reboot.  

Same rules apply;  if you have problems afterwards, the first thing you should try is a reboot (especially if you are using a debian-derived kernel.) 

edit:  at this point, most of you should be back up, and the rest will follow shortly.  

edit: everyone is back.  complain if you are still broken.  we'll be rebuilding the raid shortly
this one is on a modern server with AHCI sata, so no reboot.  it's still gotta rebuild the raid, though, and that does impact performance, even on a raid 1+0

edit: 
the raid rebuild is now done;  performance should be back to normal.  

rebooting branch to add new disks

| | Comments (0)
yes, we still have some socket F / nvidia motherboards in production.   branch needs a reboot to see the new drives, because the nvidia sata driver sucks.    

We're using the 'xm save' capabilities, so for most of you, you will see 10-20 minutes of being offline, then you will be back, without experiencing a reboot.  

for those running netbsd or debian derived distros, if you see problems, do a reboot.  Debian5 xen kernels tend to recover poorly from a xm restore.  

We will attempt to go though people's console on startup and catch these problems for you, but just in case, all of the restore-related badness can be fixed with a reboot.

Again, most of you shouldn't have to worry about this; it should just work.  

edit: we're running some final checks, but you should all be back now.

everything looks good.  I'm rebuilding the raid.  expect poor disk performance for the next 900 minutes or so.  

edit:  the raid rebuild is done;  everything should be back to normal.  let us know if there are problems.