| Age | Commit message (Collapse) | Author |
|---|
|
|
|
ajacoutot@ ok.
|
|
ok gilles@
|
|
ok gilles@
|
|
ok gilles@
|
|
ok gilles@
|
|
ok espie
|
|
ok gilles@
|
|
ok gilles@
|
|
segv on lines consisting exclusively of whitespace.
ok gilles@
|
|
place, regardless of number of preexisting aliases.
Don't call db->sync; it's covered by db->close shortly before rename.
ok chl@ gilles@
|
|
for other reasons.
Never exit outside main; we need to return to main to clean the
temp file.
Check parse_entry return code; otherwise $? == 0 even when invalid
entries were found.
ok gilles@
|
|
Do chmod before rename to eliminate short time window when aliases.db
is "official" but not readable.
Use PATH_ALIASESDB instead of hardcoding "/etc/mail/.."
Check db->close return code.
ok gilles@ henning@
|
|
- Drop static qualifiers.
- Drop unused -d switch.
- Use 0/1 exit codes instead of <sysexits.h> defines.
ok gilles@
|
|
number of arguments.
ok gilles@
|
|
|
|
track multipath routes. Code is mostly stolen from similar fixes to ospfd but
a bit different as this kroute.c implementation is more complex because of
additional task (nexthop verification for example).
This change and the similar ospfd change were desperatly needed on larger
complex network setups which tended to end up in situation where the kernel
had a different oppinion about routing than the daemon. Now the situation
should be a lot better.
Tested and OK sthen@ gollo@
|
|
|
|
makes to code simpler. While the routing table is somewhat simpler the
redistribution code does not allow a much more radical approach. With some
additional help of Mr. Kernel and his rtsock it should be possible to rip
even more code out of kroute.c.
Tested and OK sthen@, gollo@
|
|
sockaddrs included in the message. This makes the code more robust when
kernel and userland are out of sync. put it in deraadt@
|
|
ok gilles@
|
|
ok gilles@
|
|
Per RFC4271 the partial bit must be set if the attribute was not generated
on the originating router. This is mostly cosmetics.
OK sthen@
|
|
attributes and use the correct lenght in some other error cases.
OK sthen@
|
|
|
|
|
|
|
|
|
|
- makemap and newaliases need util.c now
|
|
|
|
|
|
of luck under load. Long story made short: the runner process opens
the queue and sequentially opens each bucket to process messages.
If a message is delivered by MDA/MTA after the opendir(), then the
queue process will garbage collect the message from the queue and
the runner will attempt to opendir() a path that no longer exists.
Reported by Daniel Lidberg <daniel.lidberg@gmail.com>, observed by
jacekm@ and debugged by me, that's collaborative work ;-)
|
|
|
|
|
|
encoding error or a truncation the same way. This will turn many of
our snprintf() checks into boolean checks.
|
|
Try really hard to avoid copying a file on itself: by first checking the
stat() results, and if they are not available (since some people use fucked-up
stuff like AFS), rely on the file names as a last resort.
This should avoid stuff getting copied from PKG_CACHE to PKG_CACHE and
erasing itself in a few cases...
|
|
displayed in "show rib detail".
ok claudio@
|
|
too late, ie. at alias resolution time, and it's only a log_info.
Move the check to parse.y, and make daemon die if m_src != S_DB.
ok gilles@
|
|
to start the for loop and not 1 which was correct long long time ago.
OK otto@ found by Anirban Sinha ASinha(at)zeugmasystems.com
|
|
|
|
hrProcessorLoad (where load means the percentage of time the system was
not idle during the last minute). based on HOST-RESOURCES-MIB.
ok dlg@
|
|
instead of flushing the complete source tracking table (sticky
addresses) in pf on host state changes, just flush the entries for
hosts that have been marked as down in the relayd table. this fixes
ugly problems with users loosing their sessions if another host or
redirection was going down.
ok cloder@
|
|
adds unneeded complexity. Move any stuff to rt_node structure.
|
|
ok gilles@
|
|
This is for consistency, code is not reached yet.
ok gilles@
|
|
which additionally suffixes <CRLF>, and enables EV_WRITE.
Remove bufferevent_enable(.., EV_WRITE) from session_command and
session_pickup so that EV_WRITE is enabled in exactly one place,
session_respond.
Change some responses slightly to make code fit 80 columns.
ok gilles@
|
|
Cluestick courtesy of deraadt@, discussed with todd@, mbalmer@, ok ray@.
|
|
u_int8_t to hold the value leads to invalid checking in
runner_batch_resolved(), this lead to a crash in MTA because we
assumed a batch had its mx resolved when it had not. while at it,
be more strict about errors we don't know and fatal(), it should
not happen.
ok jacekm@, ok chl@
|
|
|
|
|
