|
 |
df98bb |
From 8bdc512d2651b4600f7e744b06633a7524b64346 Mon Sep 17 00:00:00 2001
|
|
|
df98bb |
From: Lennart Poettering <lennart@poettering.net>
|
|
|
df98bb |
Date: Mon, 12 Oct 2020 16:31:42 +0200
|
|
|
df98bb |
Subject: [PATCH] sysctl: set kernel.core_pipe_limit=16
|
|
|
df98bb |
|
|
|
df98bb |
We need to make sure that our coredump pattern handler manages to read
|
|
|
df98bb |
process metadata from /proc/$PID/ before the kernel reaps the crashed
|
|
|
df98bb |
process. By default the kernel will reap the process as soon as it can.
|
|
|
df98bb |
By setting kernel.core_pipe_limit to a non-zero the kernel will wait for
|
|
|
df98bb |
userspace to finish before reaping.
|
|
|
df98bb |
|
|
|
df98bb |
We'll set the value to 16, which allows 16 crashes to be
|
|
|
df98bb |
processed in parallel. This matches the MaxConnections= setting in
|
|
|
df98bb |
systemd-coredump.socket.
|
|
|
df98bb |
|
|
|
df98bb |
See: #17301
|
|
|
df98bb |
|
|
|
df98bb |
(This doesn't close 17301, since we probably should also gracefully
|
|
|
df98bb |
handle if /proc/$PID/ vanished already while our coredump handler runs,
|
|
|
df98bb |
just in case people loclly set the sysctl back to zero. i.e. we should
|
|
|
df98bb |
collect what we can and rather issue an incomplete log record than
|
|
|
df98bb |
none.)
|
|
|
df98bb |
|
|
|
df98bb |
(cherry picked from commit 2a9b9323cd844baae3229e9dba67e478bee70654)
|
|
|
df98bb |
|
|
|
df98bb |
Resolves: #1949729
|
|
|
df98bb |
---
|
|
|
df98bb |
sysctl.d/50-coredump.conf.in | 11 +++++++++++
|
|
|
df98bb |
1 file changed, 11 insertions(+)
|
|
|
df98bb |
|
|
|
df98bb |
diff --git a/sysctl.d/50-coredump.conf.in b/sysctl.d/50-coredump.conf.in
|
|
|
df98bb |
index ccd5c2cc56..8d6fbb718c 100644
|
|
|
df98bb |
--- a/sysctl.d/50-coredump.conf.in
|
|
|
df98bb |
+++ b/sysctl.d/50-coredump.conf.in
|
|
|
df98bb |
@@ -10,3 +10,14 @@
|
|
|
df98bb |
# setting below.
|
|
|
df98bb |
|
|
|
df98bb |
kernel.core_pattern=|@rootlibexecdir@/systemd-coredump %P %u %g %s %t %c %h %e
|
|
|
df98bb |
+
|
|
|
df98bb |
+# Allow that 16 coredumps are dispatched in parallel by the kernel. We want to
|
|
|
df98bb |
+# be able to collect process metadata from /proc/%P/ while processing
|
|
|
df98bb |
+# coredumps, and thus need to make sure the crashed processes are not reaped
|
|
|
df98bb |
+# until we finished collecting what we need. The kernel default for this sysctl
|
|
|
df98bb |
+# is "0" which means the kernel doesn't wait for userspace processes to finish
|
|
|
df98bb |
+# processing before reaping the crashed processes — by setting this higher the
|
|
|
df98bb |
+# kernel will delay reaping until we are done, but only for the specified
|
|
|
df98bb |
+# number of crashes in parallel. The value of 16 is chosen to match
|
|
|
df98bb |
+# systemd-coredump.socket's MaxConnections= value.
|
|
|
df98bb |
+kernel.core_pipe_limit=16
|