[Pacemaker] solaris problem
Grüninger, Andreas (LGL Extern)
Andreas.Grueninger at lgl.bwl.de
Mon Mar 25 12:35:37 EDT 2013
Andrei
There is no need to make this change.
I described in
http://grueni.github.com/libqb/
how I compiled libqb and the other programs.
LOCALSTATEDIR should be defined with ./configure.
Please look a "Compile Corosync" in my description.
I guess your start scripts should be changed.
We use this as start script called by the smf instance
######################
#!/usr/bin/bash
# Start/stop HACluster service
#
. /lib/svc/share/smf_include.sh
## Tracing mit debug version
# PCMK_trace_files=1
# PCMK_trace_functions=1
# PCMK_trace_formats=1
# PCMK_trace_tags=1
export PCMK_ipc_type=socket
CLUSTER_USER=hacluster
COROSYNC=corosync
PACEMAKERD=pacemakerd
PACEMAKER_PROCESSES=pacemaker
APPPATH=/opt/ha/sbin/
SLEEPINTERVALL=10
SLEEPCOUNT=5
SLEPT=0
killapp() {
pid=`pgrep -f $1`
if [ "x$pid" != "x" ]; then
kill -9 $pid
fi
return 0
}
start0() {
stop0
su ${CLUSTER_USER} -c ${APPPATH}${COROSYNC}
sleep $sleep0
su ${CLUSTER_USER} -c ${APPPATH}${PACEMAKERD} &
return 0
}
stop0() {
# first try, graceful shutdown
pid=`pgrep -U ${CLUSTER_USER} -f ${PACEMAKERD}`
if [ "x$pid" != "x" ]; then
${APPPATH}${PACEMAKERD} --shutdown
sleep $SLEEPINTERVALL
fi
# second try, kill the rest
killapp ${APPPATH}${COROSYNC}
killapp ${PACEMAKER_PROCESSES}
return 0
}
let sleep0=$SLEEPINTERVALL/2
case "$1" in
'start')
start0
;;
'restart')
stop0
start0
;;
'stop')
stop0
;;
*)
echo "Usage: -bash { start | stop | restart}"
exit 1
;;
esac
exit 0
###############################
Andreas
-----Ursprüngliche Nachricht-----
Von: Andrei Belov [mailto:defanator at gmail.com]
Gesendet: Montag, 25. März 2013 15:08
An: The Pacemaker cluster resource manager
Betreff: Re: [Pacemaker] solaris problem
Ok, I fixed this issue with the following patch against libqb 0.14.4:
--- lib/unix.c.orig 2013-03-25 12:30:50.445762231 +0000
+++ lib/unix.c 2013-03-25 12:49:59.322276376 +0000
@@ -83,7 +83,7 @@
#if defined(QB_LINUX) || defined(QB_CYGWIN)
snprintf(path, PATH_MAX, "/dev/shm/%s", file); #else
- snprintf(path, PATH_MAX, LOCALSTATEDIR "/run/%s", file);
+ snprintf(path, PATH_MAX, "%s/%s", SOCKETDIR, file);
is_absolute = path;
#endif
}
@@ -91,7 +91,7 @@
if (fd < 0 && !is_absolute) {
qb_util_perror(LOG_ERR, "couldn't open file %s", path);
- snprintf(path, PATH_MAX, LOCALSTATEDIR "/run/%s", file);
+ snprintf(path, PATH_MAX, "%s/%s", SOCKETDIR, file);
fd = open_mmap_file(path, file_flags);
if (fd < 0) {
res = -errno;
libqb was configured with --with-socket-dir=/var/run/qb, /var/run/qb owned by hacluster:haclient - this configuration works fine with both corosync 2.3.0 and pacemaker 1.1.8.
Though I'm not sure that libqb is the right place to touch - maybe it'd be better to add some enhancements to pacemaker's lib/common/mainloop.c,
mainloop_add_ipc_server() ?
Cheers.
On Mar 25, 2013, at 16:01 , Andrei Belov <defanator at gmail.com> wrote:
>
> I've rebuilt libqb using separated SOCKETDIR (/var/run/qb), and set hacluster:haclient ownership to this dir.
>
> After that pacemakerd has been successfully started with all its childs:
>
> [root at ha1 /var/run/qb]# pacemakerd -fV Could not establish pacemakerd
> connection: Connection refused (146)
> info: crm_ipc_connect: Could not establish pacemakerd connection: Connection refused (146)
> info: get_cluster_type: Detected an active 'corosync' cluster
> info: read_config: Reading configure for stack: corosync
> notice: crm_add_logfile: Additional logging available in /var/log/cluster/corosync.log
> notice: main: Starting Pacemaker 1.1.8 (Build: 1f8858c): ncurses libqb-logging libqb-ipc upstart systemd corosync-native
> info: main: Maximum core file size is: 18446744073709551613
> info: qb_ipcs_us_publish: server name: pacemakerd
> notice: update_node_processes: 48de70 Node 182452614 now known as ha1, was:
> info: start_child: Forked child 60719 for process cib
> info: start_child: Forked child 60720 for process stonith-ng
> info: start_child: Forked child 60721 for process lrmd
> info: start_child: Forked child 60722 for process attrd
> info: start_child: Forked child 60723 for process pengine
> info: start_child: Forked child 60724 for process crmd
> info: main: Starting mainloop
>
> [root at ha1 /var/run/qb]# ls -l
> total 0
> srwxrwxrwx 1 hacluster root 0 Mar 25 11:50 attrd
> srwxrwxrwx 1 root root 0 Mar 25 11:43 cfg
> srwxrwxrwx 1 hacluster root 0 Mar 25 11:50 cib_ro srwxrwxrwx 1
> hacluster root 0 Mar 25 11:50 cib_rw srwxrwxrwx 1 hacluster root 0 Mar
> 25 11:50 cib_shm
> srwxrwxrwx 1 root root 0 Mar 25 11:43 cmap
> srwxrwxrwx 1 root root 0 Mar 25 11:43 cpg
> srwxrwxrwx 1 root root 0 Mar 25 11:50 lrmd
> srwxrwxrwx 1 root root 0 Mar 25 11:50 pacemakerd
> srwxrwxrwx 1 hacluster root 0 Mar 25 11:50 pengine
> srwxrwxrwx 1 root root 0 Mar 25 11:43 quorum
> srwxrwxrwx 1 root root 0 Mar 25 11:50 stonith-ng
>
> However, libqb still can not create some files in /var/run due to insufficient permissions:
>
> Mar 25 11:50:45 [60719] cib: info: init_cs_connection_once: Connection to 'corosync': established
> Mar 25 11:50:45 [60719] cib: info: crm_get_peer: Node 182452614 is now known as ha1
> Mar 25 11:50:45 [60719] cib: info: crm_get_peer: Node 182452614 has uuid 182452614
> Mar 25 11:50:45 [60719] cib: info: qb_ipcs_us_publish: server name: cib_ro
> Mar 25 11:50:45 [60719] cib: info: qb_ipcs_us_publish: server name: cib_rw
> Mar 25 11:50:45 [60719] cib: info: qb_ipcs_us_publish: server name: cib_shm
> Mar 25 11:50:45 [60719] cib: info: cib_init: Starting cib mainloop
> Mar 25 11:50:45 [60719] cib: info: pcmk_cpg_membership: Joined[0.0] cib.182452614
> Mar 25 11:50:45 [60719] cib: info: pcmk_cpg_membership: Member[0.0] cib.182452614
> Mar 25 11:50:45 [60719] cib: info: pcmk_cpg_membership: Member[0.1] cib.182452614
> Mar 25 11:50:46 [60719] cib: error: qb_sys_mmap_file_open: couldn't open file /var/run/qb-cib_rw-control-60719-60720-15: Permission denied (13)
> Mar 25 11:50:46 [60719] cib: error: qb_ipcs_us_connect: couldn't create file for mmap (60719-60720-15): Permission denied (13)
> Mar 25 11:50:46 [60719] cib: error: handle_new_connection: Invalid IPC credentials (60719-60720-15).
> Mar 25 11:50:46 [60720] stonith-ng: info: crm_ipc_connect: Could not establish cib_rw connection: Permission denied (13)
> Mar 25 11:50:46 [60719] cib: error: qb_sys_mmap_file_open: couldn't open file /var/run/qb-cib_shm-control-60719-60724-16: Permission denied (13)
> Mar 25 11:50:46 [60719] cib: error: qb_ipcs_us_connect: couldn't create file for mmap (60719-60724-16): Permission denied (13)
> Mar 25 11:50:46 [60719] cib: error: handle_new_connection: Invalid IPC credentials (60719-60724-16).
> Mar 25 11:50:46 [60724] crmd: info: crm_ipc_connect: Could not establish cib_shm connection: Permission denied (13)
> Mar 25 11:50:46 [60724] crmd: info: do_cib_control: Could not connect to the CIB service: Transport endpoint is not connected
> Mar 25 11:50:46 [60724] crmd: warning: do_cib_control: Couldn't complete CIB registration 1 times... pause and retry
>
>
> If someone has working setup on Linux with corosync 2.x, libqb and pacemaker 1.1.x - I'd be very appreciated for sharing some information about a places which libqb uses for its special socket files.
>
> Thanks in advance!
>
> (Can we say now that this problem is libqb-related, not pacemaker?)
>
>
>
> On Mar 25, 2013, at 15:30 , Andrei Belov <defanator at gmail.com> wrote:
>
>> Andreas,
>>
>> just tried "PCMK_ipc_type=socket pacemaker -fV" - a bunch of additional "event_send" errors appeared:
>>
>> Mar 25 11:15:55 [33641] ha1 corosync error [MAIN ] event_send retuned -32, expected 256!
>> Mar 25 11:15:55 [33641] ha1 corosync error [SERV ] event_send retuned -32, expected 217!
>> Mar 25 11:15:55 [33641] ha1 corosync error [SERV ] event_send retuned -32, expected 219!
>> Mar 25 11:15:55 [33641] ha1 corosync error [SERV ] event_send retuned -32, expected 223!
>> Mar 25 11:15:55 [33641] ha1 corosync error [SERV ] event_send retuned -32, expected 223!
>> Mar 25 11:15:55 [33641] ha1 corosync error [SERV ] event_send retuned -32, expected 223!
>> Mar 25 11:15:55 [33641] ha1 corosync error [SERV ] event_send retuned -32, expected 223!
>> Mar 25 11:15:55 [33641] ha1 corosync error [SERV ] event_send retuned -32, expected 223!
>> Mar 25 11:15:55 [33641] ha1 corosync error [SERV ] event_send retuned -32, expected 223!
>> Mar 25 11:15:55 [33641] ha1 corosync error [SERV ] event_send retuned -32, expected 256!
>> Mar 25 11:15:55 [53980] pengine: error: qb_ipcs_us_publish: Could not bind AF_UNIX (/var/run/pengine): Permission denied (13)
>> Mar 25 11:15:55 [53980] pengine: error: mainloop_add_ipc_server: Could not start pengine IPC server: Unknown error (-13)
>> Mar 25 11:15:55 [53980] pengine: error: main: Couldn't start IPC server
>> Mar 25 11:15:55 [53975] pacemakerd: error: pcmk_child_exit: Child process pengine exited (pid=53980, rc=1)
>> Mar 25 11:15:55 [33641] ha1 corosync error [SERV ] event_send retuned -32, expected 256!
>> Mar 25 11:15:55 [33641] ha1 corosync error [SERV ] event_send retuned -32, expected 223!
>> Mar 25 11:15:55 [53979] attrd: error: qb_ipcs_us_publish: Could not bind AF_UNIX (/var/run/attrd): Permission denied (13)
>> Mar 25 11:15:55 [53979] attrd: error: mainloop_add_ipc_server: Could not start attrd IPC server: Unknown error (-13)
>> Mar 25 11:15:55 [53979] attrd: error: main: Could not start IPC server
>> Mar 25 11:15:55 [53979] attrd: error: main: Aborting startup
>> Mar 25 11:15:55 [33641] ha1 corosync error [SERV ] event_send retuned -32, expected 223!
>> Mar 25 11:15:55 [53975] pacemakerd: error: pcmk_child_exit: Child process attrd exited (pid=53979, rc=100)
>> Mar 25 11:15:55 [33641] ha1 corosync error [SERV ] event_send retuned -32, expected 223!
>> Mar 25 11:15:55 [33641] ha1 corosync error [SERV ] event_send retuned -32, expected 223!
>> Mar 25 11:15:55 [33641] ha1 corosync error [SERV ] event_send retuned -32, expected 223!
>> Mar 25 11:15:55 [33641] ha1 corosync error [SERV ] event_send retuned -32, expected 223!
>> Mar 25 11:15:55 [33641] ha1 corosync error [SERV ] event_send retuned -32, expected 256!
>> Mar 25 11:15:55 [53976] cib: error: qb_ipcs_us_publish: Could not bind AF_UNIX (/var/run/cib_ro): Permission denied (13)
>> Mar 25 11:15:55 [53976] cib: error: mainloop_add_ipc_server: Could not start cib_ro IPC server: Unknown error (-13)
>> Mar 25 11:15:55 [53976] cib: error: qb_ipcs_us_publish: Could not bind AF_UNIX (/var/run/cib_rw): Permission denied (13)
>> Mar 25 11:15:55 [53976] cib: error: mainloop_add_ipc_server: Could not start cib_rw IPC server: Unknown error (-13)
>> Mar 25 11:15:55 [53976] cib: error: qb_ipcs_us_publish: Could not bind AF_UNIX (/var/run/cib_shm): Permission denied (13)
>> Mar 25 11:15:55 [53976] cib: error: mainloop_add_ipc_server: Could not start cib_shm IPC server: Unknown error (-13)
>> Mar 25 11:15:55 [53976] cib: error: cib_init: Couldnt start all IPC channels, exiting.
>> Mar 25 11:15:55 [53975] pacemakerd: error: pcmk_child_exit: Child process cib exited (pid=53976, rc=255)
>> Mar 25 11:15:55 [33641] ha1 corosync error [SERV ] event_send retuned -32, expected 223!
>> Mar 25 11:16:04 [53977] stonith-ng: error: setup_cib: Could not connect to the CIB service: -134 fffffd7fc421a0b0
>> Mar 25 11:16:04 [33641] ha1 corosync error [SERV ] event_send retuned -32, expected 217!
>> Mar 25 11:16:04 [53975] pacemakerd: notice: pcmk_shutdown_worker: Attempting to inhibit respawning after fatal error
>>
>>
>> # fgrep 32 /usr/include/sys/errno.h
>> #define EPIPE 32 /* Broken pipe */
>>
>>
>>
>> On Mar 25, 2013, at 13:55 , "Grüninger, Andreas (LGL Extern)" <Andreas.Grueninger at lgl.bwl.de> wrote:
>>
>>> With solaris/openindiana you should use this setting export
>>> PCMK_ipc_type=socket
>>>
>>> Andreas
>>>
>>> -----Ursprüngliche Nachricht-----
>>> Von: Andrei Belov [mailto:defanator at gmail.com]
>>> Gesendet: Montag, 25. März 2013 10:43
>>> An: pacemaker at oss.clusterlabs.org
>>> Betreff: [Pacemaker] solaris problem
>>>
>>> Hi folks,
>>>
>>> I'm trying to build test HA cluster on Solaris 5.11 using libqb 0.14.4, corosync 2.3.0 and pacemaker 1.1.8, and I'm facing a strange problem while starting pacemaker.
>>>
>>> Log shows the following errors:
>>>
>>> Mar 25 09:21:26 [33720] lrmd: error: mainloop_add_ipc_server: Could not start lrmd IPC server: Unknown error (-48)
>>> Mar 25 09:21:26 [33720] lrmd: error: try_server_create: New IPC server could not be created because another lrmd process exists, sending shutdown command to old lrmd process.
>>> Mar 25 09:21:26 [33720] lrmd: error: mainloop_add_ipc_server: Could not start lrmd IPC server: Unknown error (-48)
>>> Mar 25 09:21:26 [33720] lrmd: error: try_server_create: New IPC server could not be created because another lrmd process exists, sending shutdown command to old lrmd process.
>>> Mar 25 09:21:26 [33720] lrmd: error: mainloop_add_ipc_server: Could not start lrmd IPC server: Unknown error (-48)
>>> Mar 25 09:21:26 [33720] lrmd: error: try_server_create: New IPC server could not be created because another lrmd process exists, sending shutdown command to old lrmd process.
>>> Mar 25 09:21:26 [33720] lrmd: error: mainloop_add_ipc_server: Could not start lrmd IPC server: Unknown error (-48)
>>> Mar 25 09:21:26 [33720] lrmd: error: try_server_create: New IPC server could not be created because another lrmd process exists, sending shutdown command to old lrmd process.
>>> Mar 25 09:21:26 [33720] lrmd: error: mainloop_add_ipc_server: Could not start lrmd IPC server: Unknown error (-48)
>>> Mar 25 09:21:26 [33720] lrmd: error: try_server_create: New IPC server could not be created because another lrmd process exists, sending shutdown command to old lrmd process.
>>> Mar 25 09:21:26 [33720] lrmd: error: mainloop_add_ipc_server: Could not start lrmd IPC server: Unknown error (-48)
>>> Mar 25 09:21:26 [33720] lrmd: error: try_server_create: New IPC server could not be created because another lrmd process exists, sending shutdown command to old lrmd process.
>>> Mar 25 09:21:26 [33720] lrmd: error: mainloop_add_ipc_server: Could not start lrmd IPC server: Unknown error (-48)
>>> Mar 25 09:21:26 [33720] lrmd: error: try_server_create: New IPC server could not be created because another lrmd process exists, sending shutdown command to old lrmd process.
>>> Mar 25 09:21:26 [33720] lrmd: error: mainloop_add_ipc_server: Could not start lrmd IPC server: Unknown error (-48)
>>> Mar 25 09:21:26 [33720] lrmd: error: try_server_create: New IPC server could not be created because another lrmd process exists, sending shutdown command to old lrmd process.
>>> Mar 25 09:21:26 [33720] lrmd: error: mainloop_add_ipc_server: Could not start lrmd IPC server: Unknown error (-48)
>>> Mar 25 09:21:26 [33720] lrmd: error: try_server_create: New IPC server could not be created because another lrmd process exists, sending shutdown command to old lrmd process.
>>> Mar 25 09:21:26 [33720] lrmd: error: mainloop_add_ipc_server: Could not start lrmd IPC server: Unknown error (-48)
>>> Mar 25 09:21:26 [33720] lrmd: error: try_server_create: New IPC server could not be created because another lrmd process exists, sending shutdown command to old lrmd process.
>>> Mar 25 09:21:26 [33720] lrmd: error: main: Failed to allocate lrmd server. shutting down
>>> Mar 25 09:21:26 [33722] pengine: error: mainloop_add_ipc_server: Could not start pengine IPC server: Unknown error (-48)
>>> Mar 25 09:21:26 [33722] pengine: error: main: Couldn't start IPC server
>>> Mar 25 09:21:26 [33717] pacemakerd: error: pcmk_child_exit: Child process lrmd exited (pid=33720, rc=255)
>>> Mar 25 09:21:26 [33721] attrd: error: qb_ipcs_us_publish: Could not bind AF_UNIX (/var/run/attrd): Permission denied (13)
>>> Mar 25 09:21:26 [33721] attrd: error: mainloop_add_ipc_server: Could not start attrd IPC server: Unknown error (-13)
>>> Mar 25 09:21:26 [33721] attrd: error: main: Could not start IPC server
>>> Mar 25 09:21:26 [33721] attrd: error: main: Aborting startup
>>> Mar 25 09:21:26 [33717] pacemakerd: error: pcmk_child_exit: Child process pengine exited (pid=33722, rc=1)
>>> Mar 25 09:21:26 [33717] pacemakerd: error: pcmk_child_exit: Child process attrd exited (pid=33721, rc=100)
>>> Mar 25 09:21:26 [33718] cib: error: qb_ipcs_us_publish: Could not bind AF_UNIX (/var/run/cib_ro): Permission denied (13)
>>> Mar 25 09:21:26 [33718] cib: error: mainloop_add_ipc_server: Could not start cib_ro IPC server: Unknown error (-13)
>>> Mar 25 09:21:26 [33718] cib: error: qb_ipcs_us_publish: Could not bind AF_UNIX (/var/run/cib_rw): Permission denied (13)
>>> Mar 25 09:21:26 [33718] cib: error: mainloop_add_ipc_server: Could not start cib_rw IPC server: Unknown error (-13)
>>> Mar 25 09:21:26 [33718] cib: error: mainloop_add_ipc_server: Could not start cib_shm IPC server: Unknown error (-48)
>>> Mar 25 09:21:26 [33718] cib: error: cib_init: Couldnt start all IPC channels, exiting.
>>> Mar 25 09:21:26 [33717] pacemakerd: error: pcmk_child_exit: Child process cib exited (pid=33718, rc=255)
>>> Mar 25 09:21:35 [33719] stonith-ng: error: setup_cib: Could not connect to the CIB service: -134 fffffd7fc421a0b0
>>> Mar 25 09:21:35 [33717] pacemakerd: notice: pcmk_shutdown_worker: Attempting to inhibit respawning after fatal error
>>>
>>> Full log (in case of any things I've probably missed) is attached.
>>>
>>> I wonder to know the reason of "unknown error (-48)" - on this system 48 in errno.h is "ENOTSUP", but I haven't found the exact place in code where this may happen (so I'm not sure about that).
>>>
>>> Just for record - I'm able to run corosync on two nodes and see them connected without any visible problems - thus, I suppose there may be something wrong with either pacemaker or libqb.
>>>
>>> Any help will be greatly appreciated!
>>>
>>> Thanks,
>>> Andrei.
>>>
>>>
>>>
>>> _______________________________________________
>>> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
>>> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>>
>>> Project Home: http://www.clusterlabs.org Getting started:
>>> http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>>> Bugs: http://bugs.clusterlabs.org
>>
>
_______________________________________________
Pacemaker mailing list: Pacemaker at oss.clusterlabs.org http://oss.clusterlabs.org/mailman/listinfo/pacemaker
Project Home: http://www.clusterlabs.org Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
Bugs: http://bugs.clusterlabs.org
More information about the Pacemaker
mailing list