On 2/16/2011 4:18 AM, William Grant wrote:
> There are actually 5 handles per connection: the extra is a unix socket
> to the forking master.
>
> Locally I cannot reproduce the suspected initial leak. However, once the
> FD limit is reached it starts leaking huge numbers of master sockets and
> a few child fifos. So an lsof afterwards would probably have had limited
> utility.
>
So is this the FD limit in the Conch process? Causing children to be
spawned but not connected to, and then just go dormant. Do we then run
out of system handles, causing the Conch process to continue this cycle?
John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On 2/16/2011 4:18 AM, William Grant wrote:
> There are actually 5 handles per connection: the extra is a unix socket
> to the forking master.
>
> Locally I cannot reproduce the suspected initial leak. However, once the
> FD limit is reached it starts leaking huge numbers of master sockets and
> a few child fifos. So an lsof afterwards would probably have had limited
> utility.
>
So is this the FD limit in the Conch process? Causing children to be
spawned but not connected to, and then just go dormant. Do we then run
out of system handles, causing the Conch process to continue this cycle?
John
=:->
-----BEGIN PGP SIGNATURE----- enigmail. mozdev. org/
b8U4ACgkQJdeBCY SNAAMB0QCgmsOzK ubvn1beGHofKQbC B+jt 9sGJtyMPNwNO+ cS8Mk13
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://
iEYEARECAAYFAk1
6SsAni0abIo+
=dKwp
-----END PGP SIGNATURE-----