Updates:
Status: Fixed
Comment #1 on issue 1046 by [email protected]: NodeSshKeyAdd does not
handle updates gracefully
https://code.google.com/p/ganeti/issues/detail?id=1046
Fixed with:
commit bacd27531526faa60e595b3b4a4498a35b258779
Author: Helga Velroyen <[email protected]>
Date: Fri Apr 17 14:35:06 2015 +0200
NodeSshRemoveKey: Add retries for updating the target node
This patch adds retries to the part of 'NodeSshRemoveKey',
where the target node itself is updated. As this is likely
to fail if the node is for example already offline, we
do not fail the entire operation, but emit a non-disruptive
error which can be showing as feedback by the LU.
Signed-off-by: Helga Velroyen <[email protected]>
Reviewed-by: Hrvoje Ribicic <[email protected]>
commit b54540a37bc97b8c6a1a358e8cfbff519692a2b0
Author: Helga Velroyen <[email protected]>
Date: Fri Apr 17 12:45:03 2015 +0200
RemoveNodeSshKey: use retries when updating other nodes
When removing an SSH key, RemoveNodeSshKey so far did only
try to contact each node once. This made the entire
procedure not very resilient towards temporary network
failures. This patch adds a couple of retries for contacting
each node and only gives up after those are exhausted.
Signed-off-by: Helga Velroyen <[email protected]>
Reviewed-by: Hrvoje Ribicic <[email protected]>
commit 6ad504bd7696e49e9f68d9e627112a5ad866dc94
Author: Helga Velroyen <[email protected]>
Date: Fri Apr 17 11:47:54 2015 +0200
AddNodeSshKey: retries for all non-master nodes
This patch introduces retries for updating non-master nodes
when a new SSH key is added to the cluster. If an SSH update
fails even after the maximum number of retries is exhausted,
this does not raise an exception, but an error message is
added to the result of the method, which can be displayed
in the LU operations so that the admin sees which nodes
were the problem and need further attention.
Signed-off-by: Helga Velroyen <[email protected]>
Reviewed-by: Hrvoje Ribicic <[email protected]>
commit 1a30e46a910d772058d081ad2f155da2d2eee81a
Author: Helga Velroyen <[email protected]>
Date: Thu Apr 16 16:55:16 2015 +0200
AddNodeSshKey: retry when target node not reachable
So far, if a SSH key of a node is added to the cluster
and updating the SSH setup of the node itself fails,
the entire operation fails quite disgraceful. This
patch introduces a retry mechanism and a proper cleanup
in case of a failure.
Note that there are more places where retries would
be a good idea. They'll be added in other patches.
Signed-off-by: Helga Velroyen <[email protected]>
Reviewed-by: Petr Pudlak <[email protected]>
commit 49f49f81148ffcde660eee58ccfdafb2010e2329
Author: Helga Velroyen <[email protected]>
Date: Thu Apr 16 15:02:47 2015 +0200
Consider offline nodes when removing SSH keys
Like the previous patch, also NodeSshRemoveKey did not
consider offline nodes so far. This patch fixes it.
This fixes issue 1047.
Signed-off-by: Helga Velroyen <[email protected]>
Reviewed-by: Petr Pudlak <[email protected]>
commit d95ba4afd730c87198cfd17ae32e5211a2723d64
Author: Helga Velroyen <[email protected]>
Date: Thu Apr 16 15:02:08 2015 +0200
Consider offline nodes in NodeSshKeyAdd
So far the backend function NodeSshKeyAdd naively assumed
that all nodes are online and reachable via SSH. This
patch adds offline status information to the function's
signature to respect the offline status when updating
SSH keys.
Signed-off-by: Helga Velroyen <[email protected]>
Reviewed-by: Petr Pudlak <[email protected]>
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings
