Clustered filesystems for mix of trusted and untrusted nodes
View Patent ↗A cluster of computer system nodes share direct read/write access to storage devices via a storage area network using a cluster filesystem. At least one trusted metadata server assigns a mandatory access control label as an extended attribute of each filesystem object regardless of whether required by a client node accessing the filesystem object. The mandatory access control label indicates the sensitivity and integrity of the filesystem object and is used by the trusted metadata server(s) to control access to the filesystem object by all client nodes.
1. A method for recovery of a failed node, the method comprising:
interrupting threads processing messages from a failed node that are waiting for tokens to be returned;
sending interrupts to one or more threads that are processing messages from the failed node and that are waiting for a token state to stabilize;
sending interrupts to one or more threads that are attempting to lend tokens and that are waiting for the token state to stabilize; and
sending interrupts to one or more threads forced offline because of an error and that are waiting for the token state to stabilize.
2. The method of claim 1 , further comprising:
determining, by the threads which have sent interrupts and that are waiting for the token state to stabilize, whether the threads are allowed to wait before being interrupted.
3. The method of claim 2 , further comprising:
waking one or more sleeping threads;
checking the token state of the awoken threads;
comparing the token state of each of the awoken threads to a previous state of the awoken thread;
determining, for each respective awoken thread, if the state of each respective awoken thread has changed from a previous state; and
setting each respective awoken thread having a state that has not changed in a sleep state.
4. The method of claim 3 , further comprising:
returning partial results to token functions for each of the respective awoken threads with an early state.
5. The method of claim 4 , further comprising:
leaving a token server data structure in a partially recalled state for the threads that are waiting for tokens to be returned.
6. The method of claim 5 , further comprising:
clearing the early state of threads that are waiting for tokens to be returned when a token corresponding to the last returning token returns.
7. The method of claim 3 , further comprising:
returning the threads attempting to lend tokens early, without waiting for all tokens that correspond to the threads attempting to lend tokens to be returned.
8. A non-transitory computer readable storage medium having embodied thereon a program executable by a processor to enable a method for recovery of a failed node, the method comprising:
interrupting threads processing messages from a failed node that are waiting for tokens to be returned;
sending interrupts to one or more threads that are processing messages from the failed node and that are waiting for a token state to stabilize;
sending interrupts to one or more threads that are attempting to lend tokens and that are waiting for the token state to stabilize; and
sending interrupts to one or more threads forced offline because of an error and that are waiting for the token state to stabilize.
9. The non-transitory computer readable storage medium of claim 8 , further comprising:
determining, by the threads which have sent interrupts and that are waiting for the token state to stabilize, whether the threads are allowed to wait before being interrupted.
10. The non-transitory computer readable storage medium of claim 9 , further comprising:
waking one or more sleeping threads;
checking the token state of the awoken threads;
comparing the token state of each of the awoken threads to a previous state of the awoken thread;
determining, for each respective awoken thread, if the state of each respective awoken thread has changed from a previous state; and
setting each respective awoken thread having a state that has not changed in a sleep state.
11. The non-transitory computer readable storage medium of claim 10 , further comprising:
returning partial results to token functions for each of the respective awoken threads with an early state.
12. The non-transitory computer readable storage medium of claim 11 , further comprising:
leaving a token server data structure in a partially recalled state for the threads that are waiting for tokens to be returned.
13. The non-transitory computer readable storage medium of claim 12 , further comprising:
clearing the early state of threads that are waiting for tokens to be returned when a token corresponding to the last returning token returns.
14. The non-transitory computer readable storage medium of claim 10 , further comprising:
returning the threads attempting to lend tokens early, without waiting for all tokens that correspond to the threads attempting to lend tokens to be returned.