Comparison to other CR projects

Revision as of 13:21, 11 July 2013 by Efiop (talk | contribs)

This pages tries to explain differences between CRIU and other C/R solutions.

DMTCP

DMTCP implements checkpoint/restore of a process on a library level. This means, that if you want to C/R some application you should launch one with DMTCP library (dynamically) linked from the very beginning. When launched like this, the DMTCP library intercepts a certain amount of library calls from the application, builds a shadow data-base of information about process' internals and then forwards the request down to glibc/kernel. The information gathered is to be used to create an image of the application. With this approach, one can only dump applications known to run successfully with the DMTCP libraries, but the latter doesn't provide proxies for all kernel APIs (for example, inotify() is known to be unsupported). Another implication of this approach is potential performance issues that arise due to proxying of requests.

Restoration of process set is also tricky, as it frequently requires restoring an object with the predefined ID and kernel is known to provide no APIs for several of them. For example, kernel cannot fork a process with the desired PID. To address that, DMTCP fools a process by intercepting the getpid() library call and providing fake PID value to the application. Such behavior is very dangerous, as application might see wrong files in the /proc filesystem if it will try to access one via its PID.

CRIU, on the other hand, doesn't require any libraries to be pre-loaded. It will checkpoint and restore any arbitrary application, as long as kernel provides all needed facilities. Kernel support for some of CRIU features were added recently, essentially meaning that a recent kernel version might be required.

CRIU, DMTCP, BLCR

“looks\seems like yes/no” - i found only unproved message(s) saying “yes”/“no”

“not yet” - it is officially planned or i found no reasons, why it can’t be done.


CRIU DMTCP BLCR
arch x86_64, ARM x86, x86_64, ARM x86,x86_64,PPC/PPC64,ARM
OS
Linux
modified kernel yes, but only for some extra features.

All unnecessary features are already in new kernel versions

no no, module can be simply modprobed


problems with installation on new kernels

special libs


no yes yes
root privileges yes, otherwise it would be unsafe,because,for example, of parasite code no no
need to modify programs no no yes

there are some difficulties with statically linked applications, and with LinuxThreads (cuz it does not support them at all)



need to prepare tasks no yes

It preloadsthe DMTCP library. That library runs before the routinemain(). It creates a second thread. Thecheckpoint thread then creates a socket to the DMTCP coordinator andregisters itself. The checkpoint thread also creates a signal handler.


yes

CR shall notify processes when a checkpoint is to occur (before the kernel takes a checkpoint) to

allow the processes to prepare itself accordingly.



Does it change behavior of the c/r-ed programs?


no yes

because of wrappers on system calls


yes

because of wrappers on system calls

migration yes

even if kernel ,libs, etc are newer


Can use Memory Changes Tracking to decrease time for dumping

yes

if both kernels are recent

yes

but if all is the same!


if even prelinked addresses are different,it will not restore


But it can save the whole used libs and localization files to restore program on the different machine

Containers yes

LXC and OpenVZ containers

looks like no

It doesn't support namespaces, so it probably can’t dump containers


looks like no
parallel/distributed computations no yes

OpenMPI, MPICH2, OpenMP, Cilk are alredy supported and Infiniband is in progress.

yes

Cray MPI, Intel MPI, LAM/MPI, MPICH-V, MPICH2, MVAPICH, Open MPI, SGI MPT

c\r gdb with debugging app no, because they are using the same interface yes no
X-Windows graphics programs (KDE, GNOME, etc) yes, by using vnc yes, by using vnc seems like no



Solutions for invocation in the custom software not yet yes

Plugins and API


not yet
unix sockets yes,all kinds yes no
udp sockets yes, both ipv4 and ipv6 not yet

developers of dmtcp had no request for this

not yet
tcp sockets yes yes not yet
remote tcp connection yes not yet

but you can write a simple DMTCP plugin that tells DMTCP how you want to reconnect on restart

no
Infiniband no not yet

developing is on the half-way

no
multithread support yes yes yes
multiprocess yes yes yes
process groups yes yes not yet
zombies yes no no
namespaces yes no no
sessions yes yes not yet
Ptraced programs no yes no
System V IPC yes yes no
memory mappings yes, all kinds yes yes, partially
protected memory yes yes yes
pipes yes yes not yet
terminals yes

only Unix98 PTYs

yes


yes
non-posix files (inotify, signalfd, eventfd, etc) yes

inotify, epoll, etc.

Yes

epoll, eventfd, signalfd are already supported and

inotify will be supported in future

looks like no
timers yes no

Any counter or timer active since the beginning of a process will consider the restarted process to be a new process.

yes
Shared resources (files, mm, etc.) yes

files, memory, etc.

yes

System V shared memory(shmget, etc.), mmap-based shared memory, shared sockets, pipes, file descriptors.

no

but it is planned to suppord shared mmap regions

block devices looks like yes looks like yes no



character devices mostly no

but /dev/null, /dev/zero, etc. are supported

mostly no

looks like null and zero are supported

mostly no

but /dev/null and

/dev/zero are supported

capture the contents of all open files yes looks like no not yet


External links