On Mon, 27 Nov 2000, Robert Milkowski wrote:
] I got a lot of lines like below... Solaris 2.6 and
]squid-2.3.STABLE4-4. Any ideas? I change max file descriptors to 4096 and
]recompile squid but it didn'y help.
]
]2000/11/27 15:54:05| comm_udp_sendto: FD 17, 156.17.1.38, port 3130: (11)
]Resource temporarily unavailable
]2000/11/27 15:54:07| comm_udp_sendto: FD 17, 153.19.109.2, port 3132: (11)
]Resource temporarily unavailable
]2000/11/27 15:54:12| comm_udp_sendto: FD 17, 157.158.4.61, port 3130: (11)
]Resource temporarily unavailable
FDs are not the cause as it seems - there are very few UDP Sockets, and
those are statically allocated during Squid startup. It looks as if the
sendto() couldn't "get rid" of the PDU.
Hmm, errno is EAGAIN, here from "man sendto" on a 2.6 host:
[...]
If space is not available at the sending socket
to hold the message to be transmitted and the socket file
descriptor does have O_NONBLOCK set, sendto() will fail.
[...]
EAGAIN The socket's file descriptor is marked
O_NONBLOCK and the requested operation would
block.
Since Solaris defaults to 8KB sending UDP buffer size - which can always
only hold up to one (1) PDU (no regular buffering when sending UDP!!!),
at what rate does your Squid have to answer ICP? 50/s, 100/s?
I don't know, if udp_comm_sendto does have a limited-retry-on-EAGAIN loop
around sendto... Henrik?
Ciao,
Dipl.-Ing. Jens-S. Vöckler (voeckler@rvs.uni-hannover.de)
Institute for Computer Networks and Distributed Systems (RVS)
University of Hanover, Germany; ++49 511 762 4726
-- To unsubscribe, see http://www.squid-cache.org/mailing-lists.htmlReceived on Mon Nov 27 2000 - 08:53:16 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:56:36 MST