On 29/04/2016 23:37, Roland Kammerer wrote:
On Fri, Apr 29, 2016 at 09:36:12PM +1000, Adam Goryachev wrote:
On 29/04/2016 18:40, Roland Kammerer wrote:
On Fri, Apr 29, 2016 at 09:21:17AM +1000, Adam Goryachev wrote:
When upgrading from python-drbdmanage 0.94-1 to 0.95-1 using the packages
available at:
http://oss.linbit.com/drbdmanage/packages/drbdmanage-0.95/debian-jessie/
The following is the output:
dpkg -i python-drbdmanage_0.95-1_all.deb
(Reading database ... 48347 files and directories currently installed.)
Preparing to unpack python-drbdmanage_0.95-1_all.deb ...
Unpacking python-drbdmanage (0.95-1) over (0.94-1) ...
Setting up python-drbdmanage (0.95-1) ...
Job for drbdmanaged.socket failed. See 'systemctl status drbdmanaged.socket'
and 'journalctl -xn' for details.
Processing triggers for dbus (1.8.20-0+deb8u1) ...
Processing triggers for man-db (2.7.0.2-5) ...
systemctl status drbdmanaged.socket
● drbdmanaged.socket - DRBDManage Service
Loaded: loaded (/lib/systemd/system/drbdmanaged.socket; enabled)
Active: inactive (dead) since Fri 2016-04-29 09:15:04 AEST; 46s ago
Listen: [::]:6996 (Stream)
drbdmanage still shows that it is running and working (presumably the old
version). Doing a reboot solves the problem, but I would think the package
upgrade should handle this more smoothly. Any suggestions or ideas?
Is this a satellite node or a control node?
Control node.
Then you don't need socket activation at all. systemd disable/stop it.
Can you confirm which part/command I should be running?
Here is another machine (in the same "farm"):
root@san3:/usr/src/drbd9# systemctl |grep -i drbd
sys-devices-virtual-block-drbd0.device loaded active plugged
/sys/devices/virtual/block/drbd0
sys-devices-virtual-block-drbd1.device loaded active plugged
/sys/devices/virtual/block/drbd1
● drbd.service loaded failed failed LSB: Control drbd resources.
drbdmanaged.socket loaded active listening DRBDManage Service
This machine is still running python-drbdmanage 0.94-1 (ie, I haven't
tried to upgrade it yet).
It seems to suggest the socket (that you say is not needed on a control
node, I'm guessing it is needed on a satellite node) is working
normally. However, the service is not working.
Right now:
root@san3:/usr/src/drbd9# drbdmanage list-nodes
+------------------------------------------------------------------------------------------------------------+
| Name | Pool Size | Pool Free
| | State |
|------------------------------------------------------------------------------------------------------------|
| castle | 5403308 | 4017368
| | ok |
| san2.websitemanagers.com.au | 3777040 | 2432060
| | ok |
| san3 | 1830932 | 1830924
| | ok |
| san5 | 2764172 | 2745084
| | ok |
| xen6 | 0 | 0
| | satellite node, no storage |
+------------------------------------------------------------------------------------------------------------+
So, it seems to be working, but there are definitely some problems going
on which I'll detail in my next post (since I think they are unrelated).
I repeated the same upgrade on a second control node, with identical
results.
How was that enabled at all in the first place? By you or was it picked
up "magically" by systemd?
I didn't do anything more than install the python-drbdmanage package, so
either the package did something to make it work, or else systemd
"magically" picked it up.
Hm, should not happen, but I will check that in a clean environment.
Just to be sure about the test case:
- installed 0.94
- played around with it
- upgraded to 0.95
- *never ever* touched any sytemd foo manually
correct?
Yes, I don't really know anything about systemd, except it is default on
newer debian, and adds yet another thing I will need to learn about,
instead of the older and simpler scripts + symlinks....
So, I've definitely not manually enabled or otherwise touched the
systemd configs.
Regards,
Adam
_______________________________________________
drbd-user mailing list
[email protected]
http://lists.linbit.com/mailman/listinfo/drbd-user
