Hawkin
/
arangodb
mirror of https://gitee.com/bigwinds/arangodb
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

add some bits about linux coredumps

This commit is contained in:
Wilfried Goesgens 2017-04-07 10:43:02 +02:00
parent 2da0e41466
commit 6bb187e88a
1 changed files with 30 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

32
README_maintainers.md
View File

@ -342,9 +342,21 @@ arangod commandline arguments
__________________________________________________________________________________________________________ __________________________________________________________________________________________________________
Linux Cordeumps Linux Coredumps
=============== ===============
Hint: on Ubuntu the `apport` package may interfere with this. Generally coredumps have to be enabled using:
ulimit -c unlimited
You should then see:
ulimit -a
core file size (blocks, -c) unlimited
for each shell and its subsequent processes.
Hint: on Ubuntu the `apport` package may interfere with this; however you may use the `systemd-coredump` package
which automates much of the following:
So that the unit testing framework can autorun gdb it needs to reliably find the corefiles. So that the unit testing framework can autorun gdb it needs to reliably find the corefiles.
In Linux this is configured via the `/proc` filesystem, you can make this reboot permanent by In Linux this is configured via the `/proc` filesystem, you can make this reboot permanent by
@ -365,6 +377,22 @@ Solaris Coredumps
Solaris configures the system corefile behaviour via the `coreadm` programm. Solaris configures the system corefile behaviour via the `coreadm` programm.
see https://docs.oracle.com/cd/E19455-01/805-7229/6j6q8svhr/ for more details. see https://docs.oracle.com/cd/E19455-01/805-7229/6j6q8svhr/ for more details.
Analyzing Coredumps on Linux
============================
We offer debug packages containing the debug symbols for your binaries. Please install them if you didn't compile yourselves.
Given you saw in the log of the arangod with the PID `25216` that it died, you should then find
`/var/tmp/core-V8 WorkerThread-25216-1490887259` with this information. We may now start GDB and inspect whats going on:
gdb /usr/sbin/arangod /var/tmp/*25216*
These commands give usefull information about the incident:
backtrace full
thread apply all bt
The first gives the full stacktrace including variables of the last active thread, the later one the stacktraces of all threads.
Windows debugging Windows debugging
================= =================
For the average \*nix user windows debugging has some awkward methods. For the average \*nix user windows debugging has some awkward methods.