Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Fixed
-
10.0.24-galera
-
None
-
Redhat 6.7
Description
On a Redhat 6.7 installation we see Garbd segfaulting when we use a configuration file when starting Garbd, on Redhat 7.1 we don't see this issue happening.
[user@n1 ~]$ sudo /usr/bin/garbd --cfg /etc/garbd.cnf
|
[user@n1 ~]$ echo $?
|
139
|
Nothing gets logged to the garbd log file.
If we provide all the options inside the configuration file manually on the command line like this everything is fine:
sudo /usr/bin/garbd -g my_wsrep_cluster -a "gcomm://10.10.16.12:4567,10.10.16.13:4567" -o "gmcast.listen_addr=tcp://0.0.0.0:4567" -l /var/log/garbd.log
|
The configuration files get generated by ClusterControl for both OS-es and contain the exact same information:
[sudo@n1 ~]$ sudo cat /etc/garbd.cnf
|
address = gcomm://10.10.16.12:4567,10.10.16.13:4567
|
group = my_wsrep_cluster
|
options = gmcast.listen_addr=tcp://0.0.0.0:4567
|
log = /var/log/garbd.log
|
Also the garbd.log file exists and has the right permissions: if we start up Garbd with all the commandline parameters it is fine.
Using strace gave us the folowing last few lines:
open("/etc/garbd.cnf", O_RDONLY) = 3
|
read(3, "address=gcomm://10.10.16.12:4567"..., 8191) = 143
|
read(3, "", 8191) = 0
|
close(3) = 0
|
open("/var/log/garbd.log", O_WRONLY|O_CREAT|O_APPEND, 0666) = 3
|
fstat(3, {st_mode=S_IFREG|0644, st_size=11312, ...}) = 0
|
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7f5bcc8b9000
|
fstat(3, {st_mode=S_IFREG|0644, st_size=11312, ...}) = 0
|
lseek(3, 11312, SEEK_SET) = 11312
|
gettimeofday({1461197582, 39668}, NULL) = 0
|
open("/etc/localtime", O_RDONLY) = 4
|
fstat(4, {st_mode=S_IFREG|0644, st_size=118, ...}) = 0
|
fstat(4, {st_mode=S_IFREG|0644, st_size=118, ...}) = 0
|
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7f5bcc8b8000
|
read(4, "TZif2\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\1\0\0\0\1\0\0\0\0"..., 4096) = 118
|
lseek(4, -62, SEEK_CUR) = 56
|
read(4, "TZif2\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\1\0\0\0\1\0\0\0\0"..., 4096) = 62
|
close(4) = 0
|
munmap(0x7f5bcc8b8000, 4096) = 0
|
write(3, "2016-04-21 00:13:02.039 INFO: C"..., 69) = 69
|
--- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_MAPERR, si_addr=0xfffffffffffffff8} ---
|
+++ killed by SIGSEGV +++
|
I have added the full strace to the report.
The package we install:
galera-25.3.15-1.rhel6.el6.x86_64
Attachments
Issue Links
- relates to
-
-
- Closed
-