IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
When lvmetad is going to be shutdown - there were 2 issue:
If there was endless stream of command contacting lvmetad - the process
would never finish - so now when we recieved SIGTERM - we are no longer
accepting new connections and we just wait till the existing ones are
finished.
2nd. issue is that actually when we are waiting for finish of all client
threads - we basically want an usleep() and check if all threads
are already finished - proper solution would be to singal from a thread
there is some work to do for master thread - but that would be a bigger
change and since lvmetad is no longer developed - keep the change
minimal and just use pselect() as our '1sec.' sleeping call once
we are in 'shutdown' mode.
Reported-by: wangjufeng@huawei.com
Since we are checking _shutdown_requested - we expect all threads
are finished - that is currently checked only by checking ->next ptr
being NULL - so this can be NULL only when _reap() function clears
out all already finished threads.
I'm finding this design quite problematic in its core - but as a
'trivial hotfix' - lets _reap() linked list before check for signal.
There is likely a large potentical for few races - but the windows is
very very small - since lvmetad has been already purged from upstream,
lets go with this hotfix.
To avoid tiny race on checking arrival of signal and entering select
(that can latter remain stuck as signal was already delivered) switch
to use pselect().
If it would needed, we can eventually add extra code for older systems
without pselect(), but there are probably no such ancient systems in
use.
ATM we have several instances of daemonizing code.
Each has its 'special' logic so not completely easy
to unify them all into a single routine.
Start to unify them and use one strategy for rediricting
all input/outpus to /dev/null - use 'dup2' function for this
and open /dev/null before fork to make sure it's available.
Some archs can use even 64K pages and then lvm2 runs into trouble if
the stack is 'too small' to fit extra page capturing stack overwrite.
So when lvm2 limits stack - add extra mem page - be it 4K or 64K.
Relates to ppc64le bug: https://bugzilla.redhat.com/1387279
Currently, a shutdown signal will cause lvmetad to quit
responding to new connections, but not actually exit until
all connections are gone. If a program is maintaining a
long running connection (e.g. lvmlockd, or even an lvm
command) when lvmetad gets a shutdown signal, then all
further commands will hang indefinately waiting for a
response that won't be sent.
With this patch, make lvmetad continue handling new
connections even after a shutdown signal. It will exit
once all connections are gone.
works with systemd activated daemons only as of now
each daemon implementation may decide to signalize its
internal idle state (i.e. all background tasks unrelated to
client threads are finished)
all sockets opened by a daemon or handed over by systemd
have to have CLOEXEC flag set. Otherwise we get nasty
warnings about leaking descriptors in processes spawned by
daemon.
Ensure daemon-io.h is used as a generic header included
with configure defines before other headers.
(In future all lvm2 libraries should settle on a single lib.h header)
Rename couple defines to better match header file names.
Make it easier to run a live lvmetad in debugging mode and
to avoid conflicts if multiple test instances need to be run
alongside a live one.
No longer require -s when -f is used: use built-in default.
Add -p to lvmetad to specify the pid file.
No longer disable pidfile if -f used to run in foreground.
If specified socket file appears to be genuine but stale, remove it
before use.
On error, only remove lvmetad socket file if created by the same
process. (Previous code removes socket even while a running instance
is using it!)
For reseting locale environment into significantly less memory
consuming version 'C' - use LC_ALL instead of LANG since it has
higher priority in locale settings.
Otherwise we may observe whole locale-archive which might be
over 100MB on i.e. Fedora systems locked in memory with
some daemons.
