Michael Paterson mhpx

Organizations

@hep-gc
mhpx edited the hep-gc/cloud-scheduler wiki
@mhpx
  • @mhpx e66533c
    updated log message for clarity and changed log level. Fixed #379
mhpx commented on issue hep-gc/cloud-scheduler#378
@mhpx

Can re-use the retired_resources from reloading persistence and shuffling the extra VMs in there. That way those resources will simply get removed/…

@mhpx
  • @mhpx ed61fcd
    fixed missing argurment assignment #373
mhpx commented on issue hep-gc/cloud-scheduler#328
@mhpx

@rd37 Is this for when keystone is using https and requires a CA cert to verify the connection like in #373 or injecting another cert into a VM bei…

@mhpx
Adjust cloud resources quotas via cloud_admin without quickrestart
@mhpx
User to cloud entry mapping to restrict which users can use which clouds
mhpx commented on issue hep-gc/cloud-scheduler#305
@mhpx

looks like boto uses a cacerts.txt file located in the boto folder, would need to append any missing cert to this as the API call doesn't seem to a…

@mhpx
mhpx commented on issue hep-gc/cloud-scheduler#305
@mhpx

somewhat related to #373 May be able to use similar fix so certs can be optionally validated.

@mhpx
Unable to see spot_id on starting amazon VMs
mhpx commented on issue hep-gc/cloud-scheduler#374
@mhpx

possibly related to upgrade and changes in the cluster class.

@mhpx
possible bug reloading cs and losing slots
@mhpx
for openstack clouds using SSL/https auth specify the .pem file to use to verify
@mhpx
@mhpx
Unable to see spot_id on starting amazon VMs
@mhpx
  • @mhpx b0ee1fa
    pass over the empty partitionable slot to reduce log spam
mhpx commented on issue hep-gc/cloud-scheduler#335
@mhpx

whenever you see this happening could you do a cloud_status -x and see which thread(s) is locking up. And grep to the log for the last message from…

mhpx commented on issue hep-gc/cloud-scheduler#371
@mhpx

currently working on the ability to support mutable deployments Unless their page out of date that feature not quite there yet. Will be interesti…

mhpx commented on issue hep-gc/cloud-scheduler#371
@mhpx

well it would certainly remove having to maintain all the different cloud layers, has a rest style api to abstract all the provisioning. Would prob…

@mhpx
Retire X VMs without specifying particular ones
@mhpx
Let cloud_admin retire specific number of VMs
mhpx commented on issue hep-gc/cloud-scheduler#316
@mhpx

related / same issue - #365

@mhpx
Record reasons for jobs failing to boot vm, match cloud, etc.
@mhpx
Track which images fail to register / boot correctly
@mhpx
record flavor of vm when on openstack
@mhpx
gracefully retire instead of kill when reducing quota
mhpx commented on issue hep-gc/cloud-scheduler#352
@mhpx

related - #366 fixed - closing issue

@mhpx
Cloud scheduler destroys VMs directly after sending them the condor_off signal