Updates:
Status: Fixed
Comment #3 on issue 940 by [email protected]: Retry jobs after
ECODE_TEMP_NORES
https://code.google.com/p/ganeti/issues/detail?id=940
commit 91b8f1207b60b308d87166b25b05c93a27b36630
Author: Klaus Aehlig <[email protected]>
Date: Fri Sep 19 11:11:27 2014 +0200
Make mcpu handle retry requests internally
If the execution of a logical unit indicates that it is
temporarily out of resource (ECODE_TEMP_NORES), then make
the mcpu prepare the the logical unit for retry and restart
again.
Signed-off-by: Klaus Aehlig <[email protected]>
Reviewed-by: Petr Pudlak <[email protected]>
commit f963755896cb5edb516c7ce41981fa8f64419ff9
Author: Klaus Aehlig <[email protected]>
Date: Mon Sep 22 14:04:40 2014 +0200
Extend base LU by PrepareRetry
LUs may override this method if they know how to prepare for an internal
retry after temporarily being out of resources.
Signed-off-by: Klaus Aehlig <[email protected]>
Reviewed-by: Petr Pudlak <[email protected]>
commit edf17b5ee9e80b6a639b45e30aa57a3f317bc631
Author: Klaus Aehlig <[email protected]>
Date: Mon Sep 22 17:04:57 2014 +0200
Add new GenericErrror OpRetryNotSupportedError
LUs that temporarily run out of resources, most notably
instance creation with opportunistic locking, may opt
into retrying. If they don't do so, this is the exception
to be thrown on a retry-preparation attempt. Having a
custom exception rather than throwing NotImplementedError
pleases lint.
Signed-off-by: Klaus Aehlig <[email protected]>
Reviewed-by: Petr Pudlak <[email protected]>
commit 754dae725ec4f029725a76ce02ee2b9c66807ebd
Author: Klaus Aehlig <[email protected]>
Date: Mon Sep 22 15:50:39 2014 +0200
Reset op status to wating on NotifyRetry
When a job is retried, it is effectively reset, so we have to
reset the state back to waiting again. This will also ensure
that we're not unpleasantly surprised when NotifyStart is called
the second time.
Signed-off-by: Klaus Aehlig <[email protected]>
Reviewed-by: Petr Pudlak <[email protected]>
commit 2a71264e675a90d2f4e2aa6372a925af6f520314
Author: Klaus Aehlig <[email protected]>
Date: Mon Sep 22 15:09:19 2014 +0200
Extend the callback interface by NotifyRetry
Given that we will support internal retries after failed opportunistic
locking, also extend the callback interface accordingly.
Signed-off-by: Klaus Aehlig <[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
