128 lines · plain
1==============2Cgroup Freezer3==============4 5The cgroup freezer is useful to batch job management system which start6and stop sets of tasks in order to schedule the resources of a machine7according to the desires of a system administrator. This sort of program8is often used on HPC clusters to schedule access to the cluster as a9whole. The cgroup freezer uses cgroups to describe the set of tasks to10be started/stopped by the batch job management system. It also provides11a means to start and stop the tasks composing the job.12 13The cgroup freezer will also be useful for checkpointing running groups14of tasks. The freezer allows the checkpoint code to obtain a consistent15image of the tasks by attempting to force the tasks in a cgroup into a16quiescent state. Once the tasks are quiescent another task can17walk /proc or invoke a kernel interface to gather information about the18quiesced tasks. Checkpointed tasks can be restarted later should a19recoverable error occur. This also allows the checkpointed tasks to be20migrated between nodes in a cluster by copying the gathered information21to another node and restarting the tasks there.22 23Sequences of SIGSTOP and SIGCONT are not always sufficient for stopping24and resuming tasks in userspace. Both of these signals are observable25from within the tasks we wish to freeze. While SIGSTOP cannot be caught,26blocked, or ignored it can be seen by waiting or ptracing parent tasks.27SIGCONT is especially unsuitable since it can be caught by the task. Any28programs designed to watch for SIGSTOP and SIGCONT could be broken by29attempting to use SIGSTOP and SIGCONT to stop and resume tasks. We can30demonstrate this problem using nested bash shells::31 32 $ echo $$33 1664434 $ bash35 $ echo $$36 1669037 38 From a second, unrelated bash shell:39 $ kill -SIGSTOP 1669040 $ kill -SIGCONT 1669041 42 <at this point 16690 exits and causes 16644 to exit too>43 44This happens because bash can observe both signals and choose how it45responds to them.46 47Another example of a program which catches and responds to these48signals is gdb. In fact any program designed to use ptrace is likely to49have a problem with this method of stopping and resuming tasks.50 51In contrast, the cgroup freezer uses the kernel freezer code to52prevent the freeze/unfreeze cycle from becoming visible to the tasks53being frozen. This allows the bash example above and gdb to run as54expected.55 56The cgroup freezer is hierarchical. Freezing a cgroup freezes all57tasks belonging to the cgroup and all its descendant cgroups. Each58cgroup has its own state (self-state) and the state inherited from the59parent (parent-state). Iff both states are THAWED, the cgroup is60THAWED.61 62The following cgroupfs files are created by cgroup freezer.63 64* freezer.state: Read-write.65 66 When read, returns the effective state of the cgroup - "THAWED",67 "FREEZING" or "FROZEN". This is the combined self and parent-states.68 If any is freezing, the cgroup is freezing (FREEZING or FROZEN).69 70 FREEZING cgroup transitions into FROZEN state when all tasks71 belonging to the cgroup and its descendants become frozen. Note that72 a cgroup reverts to FREEZING from FROZEN after a new task is added73 to the cgroup or one of its descendant cgroups until the new task is74 frozen.75 76 When written, sets the self-state of the cgroup. Two values are77 allowed - "FROZEN" and "THAWED". If FROZEN is written, the cgroup,78 if not already freezing, enters FREEZING state along with all its79 descendant cgroups.80 81 If THAWED is written, the self-state of the cgroup is changed to82 THAWED. Note that the effective state may not change to THAWED if83 the parent-state is still freezing. If a cgroup's effective state84 becomes THAWED, all its descendants which are freezing because of85 the cgroup also leave the freezing state.86 87* freezer.self_freezing: Read only.88 89 Shows the self-state. 0 if the self-state is THAWED; otherwise, 1.90 This value is 1 iff the last write to freezer.state was "FROZEN".91 92* freezer.parent_freezing: Read only.93 94 Shows the parent-state. 0 if none of the cgroup's ancestors is95 frozen; otherwise, 1.96 97The root cgroup is non-freezable and the above interface files don't98exist.99 100* Examples of usage::101 102 # mkdir /sys/fs/cgroup/freezer103 # mount -t cgroup -ofreezer freezer /sys/fs/cgroup/freezer104 # mkdir /sys/fs/cgroup/freezer/0105 # echo $some_pid > /sys/fs/cgroup/freezer/0/tasks106 107to get status of the freezer subsystem::108 109 # cat /sys/fs/cgroup/freezer/0/freezer.state110 THAWED111 112to freeze all tasks in the container::113 114 # echo FROZEN > /sys/fs/cgroup/freezer/0/freezer.state115 # cat /sys/fs/cgroup/freezer/0/freezer.state116 FREEZING117 # cat /sys/fs/cgroup/freezer/0/freezer.state118 FROZEN119 120to unfreeze all tasks in the container::121 122 # echo THAWED > /sys/fs/cgroup/freezer/0/freezer.state123 # cat /sys/fs/cgroup/freezer/0/freezer.state124 THAWED125 126This is the basic mechanism which should do the right thing for user space task127in a simple scenario.128