Robert Schetterer schrieb:
> Timo Sirainen schrieb:
>> On Thu, 2008-10-02 at 20:30 +0200, Bernhard Herzog wrote:
>>> On 02.10.2008, Sascha Wilde wrote:
We discovered that this new code in dict-server.c seems to be the problem:
server->fd = net_listen_unix_unlink_stale(path, 64)
Timo Sirainen schrieb:
> On Thu, 2008-10-02 at 20:30 +0200, Bernhard Herzog wrote:
>> On 02.10.2008, Sascha Wilde wrote:
>>> We discovered that this new code in dict-server.c seems to be the problem:
>>>
>>> server->fd = net_listen_unix_unlink_stale(path, 64);
>>> if (server->fd == -1)
On Thu, 2008-10-02 at 20:30 +0200, Bernhard Herzog wrote:
> On 02.10.2008, Sascha Wilde wrote:
> > We discovered that this new code in dict-server.c seems to be the problem:
> >
> > server->fd = net_listen_unix_unlink_stale(path, 64);
> > if (server->fd == -1) {
> > if (er
On 02.10.2008, Sascha Wilde wrote:
> We discovered that this new code in dict-server.c seems to be the problem:
>
> server->fd = net_listen_unix_unlink_stale(path, 64);
> if (server->fd == -1) {
> if (errno == EADDRINUSE)
> i_fatal("Socket already exi
Hi Timo,
Hi *,
in 1.2 the dict server (tested with sqlite backend) is somewhat broken.
It bails out regularly with "Fatal: dict: Socket already exists: ..."
(looks like a race condition as it doesn't fail always).
We discovered that this new code in dict-server.c seems to be the problem:
