It is now possible to atomically persist configuration files. This feature allows users to update persisted files without overwriting the files' content and potentially breaking the system with misconfigured code.
Hello,
Currently there is no way to perform atomic store of configuration file. The persistence API of ovirtnode.ovirtfunctions can fail in many stages and leave system in unpredictable state.
I suggest an API like:
persist(file, destination)
1. cp file /config/destination.tmp
2. mv /config/destination.tmp /config/destination
3. rebind /config/destination destination
This way we have either the content of the new or old file.
---
BTW:
Looking at code, please try to do as much as you can in python. I don't understand the reason of calling system utilities such as:
system("mkdir -p /config/" + dirname)
->os.makedirs(os.path.join('/config', dirname))
system("rm -f /config"+ filename)
->shutil.rmtree(os.path.join('/config', dirname))
system_closefds("echo "+filename+" >> /config/files")
->with open('/config/files', 'a') as f:
f.write("%s\n" % filename)
ret = system_closefds("grep -q \"^$" + filename +"$\" " + \
" /config/files 2> /dev/null")
->with open('/config/files', 'r') as f:
if filename in f:
....
etc...
Using python only reduces the blocks that can fail, provide better control on code flow and much better perform.
In these small cases you must call external program, please try to avoid system() as it may route output and input of caller which is an unexpected behaviour of an API. Use subprocess.Popen().
Thank you.
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.
http://rhn.redhat.com/errata/RHBA-2013-0556.html
Hello, Currently there is no way to perform atomic store of configuration file. The persistence API of ovirtnode.ovirtfunctions can fail in many stages and leave system in unpredictable state. I suggest an API like: persist(file, destination) 1. cp file /config/destination.tmp 2. mv /config/destination.tmp /config/destination 3. rebind /config/destination destination This way we have either the content of the new or old file. --- BTW: Looking at code, please try to do as much as you can in python. I don't understand the reason of calling system utilities such as: system("mkdir -p /config/" + dirname) ->os.makedirs(os.path.join('/config', dirname)) system("rm -f /config"+ filename) ->shutil.rmtree(os.path.join('/config', dirname)) system_closefds("echo "+filename+" >> /config/files") ->with open('/config/files', 'a') as f: f.write("%s\n" % filename) ret = system_closefds("grep -q \"^$" + filename +"$\" " + \ " /config/files 2> /dev/null") ->with open('/config/files', 'r') as f: if filename in f: .... etc... Using python only reduces the blocks that can fail, provide better control on code flow and much better perform. In these small cases you must call external program, please try to avoid system() as it may route output and input of caller which is an unexpected behaviour of an API. Use subprocess.Popen(). Thank you.