The first thing that you should do is try 2006k and see if that makes the
problem go away:
ftp://ftp.cac.washington.edu/mail/imap.tar.Z
On SVR4 systems such as Solaris, it is necessary to run operations that
deal with a mailbox (copy, append, etc.) in a separate fork from the main
fork. This "execution process" is needed due to the defective-by-design
fcntl() locking mechanism that must be used on SVR4. fctnl() has certain
"last close" semantics which cause locks to be broken improperly.
The improperly broken locks, in turn, lead to mailbox corruption.
There is a communication protocol between the main process and the
execution process. The "Execution process terminated abnormally" message
is one of several fatal error conditions which can occur in this
mechanism. The particular error that you got indicates that the execution
process stopped with a SIGPIPE.
I do not know why LDAP should make a difference, but perhaps LDAP does
something with file access that impacts the execution process.
If you don't mind getting thoroughly nauseated by the contortions forced
by SVR4, the module to inspect is flocksim.c. There is a fair amount of
description text detailing the problem(s), including sample code to
demonstrate why fcntl() locking is defective by design.
This mechanism is neither needed nor used on Linux and BSD (including Mac
OS X); and thus Linux and BSD are completely immune to the problem. The
execution process code is not even compiled on those platforms. Linux and
BSD have the flock() system call which does not have the problem. Note
that even though Solaris has an flock() C library routine, that routine
calls fcntl() and thus has the problem. It is not a problem in the
interface to the call, it is a problem with its internal workings.
Hence my longstanding recommendation to get rid of Solaris and replace it
with Linux and/or BSD. This particular issue has been a headache for me
for the past 15 years.
On Thu, 1 Nov 2007, Chris Miller wrote:
Good morning.
After some time using imap 2006g without issues, I'm suddenly having an
issue with it. I'm sure this is not the fault of the imapd, but rather is
Solaris' fault (imap was not changed; Solaris was patched), but I'm hoping to
get some guidance as to what to look at.
Server: IMAP 2006g on Solaris 10
Client: PINE 4.64 on Linux (multiple flavors) and Thunderbird on Windows
Problem: Message saves to folders (both mix and mbox formats tested)
often, but not always, appear to fail with error message:
"Execution process terminated abnormally (d)". When saving
individual messages, they appear to always be successfully
saved despite the fact that pine claims the save failed.
When saving multiple messages, which almost always tickles
this error, a greater-than-1 subset of the messages is
successfully saved, but pine reports only one saved.
Here is an extract from .pine-debug1:
------ SNIP ------
----- MAIL VIEW -----
IMAP 12:04:54 10/31 mm_notify babble:
{server.example.com:993/imap/notls/ssl/user="username"}INBOX: Check
completed
IMAP 12:04:54 10/31 mm_log error: Execution process terminated abnormally
(d)
FAILED save of msg 309 (c-client sequence #)
Save to folder "mymail" FAILED
------ SNIP ------
The imapd process does not log anything at this time, and the connection
stays open and usable.
Finally, it seems the problem only occurs when the nsswitch.conf file uses
LDAP for passwd and group. If I switch these to NIS instead, the error goes
away. Presumably, this points to libsldap in some way, but I'm hoping someone
can help me figure what it could be doing to generate this particular failure
mode.
Thanks for any assistance,
Chris Miller
_______________________________________________
Imap-uw mailing list
[email protected]
https://mailman1.u.washington.edu/mailman/listinfo/imap-uw
-- Mark --
http://panda.com/mrc
Democracy is two wolves and a sheep deciding what to eat for lunch.
Liberty is a well-armed sheep contesting the vote.
_______________________________________________
Imap-uw mailing list
[email protected]
https://mailman1.u.washington.edu/mailman/listinfo/imap-uw
