Bug#841028: cups-daemon: cupsd uses 100% CPU
Dear Maintainer,
we're experiencing very similar behaviour on our end and cups-filters > 1.13 does not appear to fix the issue for us. For testing purposes we upgraded the packages on our Stretch server to their Buster counterparts:
We're currently running:
cups 2.2.1-8+deb9u1
cups-bsd 2.2.1-8+deb9u1
cups-client 2.2.1-8+deb9u1
cups-common 2.2.1-8+deb9u1
cups-core-drivers 2.2.1-8+deb9u1
cups-daemon 2.2.1-8+deb9u1
cups-filters 1.20.3-1+b1
cups-filters-core-drivers 1.20.3-1+b1
cups-ipp-utils 2.2.1-8+deb9u1
cups-pdf 2.6.1-22
cups-ppdc 2.2.1-8+deb9u1
cups-server-common 2.2.1-8+deb9u1
cupsddk 1.5.3-5+deb7u6
The cupsd process remains at 100% CPU usage.
strace (scrolling through rapidly):
poll([{fd=19, events=POLLIN}], 1, 0) = 1 ([{fd=19, revents=POLLIN}])
recv(21, "\247", 1, MSG_PEEK) = 1
poll([{fd=21, events=POLLIN}], 1, 0) = 1 ([{fd=21, revents=POLLIN}])
epoll_wait(5, [{EPOLLIN, {u32=1468342600, u64=83072721224}}, {EPOLLIN, {u32=1468196936, u64=91662510152}}], 4096, 1000) = 2
recv(19, "\303", 1, MSG_PEEK) = 1
poll([{fd=19, events=POLLIN}], 1, 0) = 1 ([{fd=19, revents=POLLIN}])
recv(21, "\247", 1, MSG_PEEK) = 1
poll([{fd=21, events=POLLIN}], 1, 0) = 1 ([{fd=21, revents=POLLIN}])
epoll_wait(5, [{EPOLLIN, {u32=1468342600, u64=83072721224}}, {EPOLLIN, {u32=1468196936, u64=91662510152}}], 4096, 1000) = 2
The file descriptors correspond to the following sockets:
cupsd 32123 root 19u IPv4 33857634 0t0 TCP iserv.dev2.iserv.eu:ipp->Dogmeat.dev2.iserv.eu:43696 (CLOSE_WAIT)
cupsd 32123 root 21u IPv4 33865820 0t0 TCP iserv.dev2.iserv.eu:ipp->Dogmeat.dev2.iserv.eu:43706 (CLOSE_WAIT)
tshark doesn't show any network traffic on these sockets.
Reply to: