one of our customers sees a 'CONTACT LOST TO ARCHIVE MANAGE' alarm during every OnLineBackup.
The alarm only exists for about a minute. Thereafter it is automatically cleared by the system.
Our customer uses Spectrum 10 in a single server environment (no fault tolerance).
The server is running on Redhat 6.7.
To check if this is a general problem with Spectrum 10 or maybe with the OS it is running on, I tried to reproduce the issue in several test labs.
In one of three environments (also Linux), I was successful. During each OLB that was executed in this test environment, the above mentioned alarm occurs as you can see in the following screenshot. However the Archive Manager was running the whole time. Thus it looks like, that this is a false alarm.
So my question would be, if someone of you made the same experience?
Is this a known bug?
Any hints would be very welcome.
Thanks and regards,
Can you please check below 2 things
Do you see any error messages in the archmgr.out or stdout..log at the sametime?
Auto start/stop Archive Manager is enabled in Spectrum Control panel?
When you run your SS backup, do you have the configuration enabled to also backup the DDM? If so, I think you would see this alarm. This behavior is configured by modifying the "post_olb_script". Though I admit, I couldn't get this alarm to replicate in me test environment, nor do we see it in production.
We have the same in a Distributed environment (alarm on 2 out of 4 SS), on Spectrum 9.4, on Linux RH6.
thanks for your hint as well, but as mentioned above, the alarm is caused due to the DDM backup option enabled within the post_olb_script.
I see no error message.
I get the alarm at:
Tue 26 Jan, 2016 - 01:32:16 - The SpectroSERVER cannotconnect to or has lost contact with the Archive Manager. -
In the logs I see:
Jan 26 01:17:16 : Closing database
Jan 26 01:17:16 : ArchMgr is shutting down...
Jan 26 01:17:16 : ArchMgr has successfully shut down.
Jan 26 01:35:50 : ArchMgr started as user 'spectrum'Jan 26 01:35:50 : ArchMgr validating database.Jan 26 01:35:50 : ArchMgr successfully connected to MySQL daemonJan 26 01:35:50 : ArchMgr loaded DDM database with landscape handle 0x3000000
Jan 26 01:35:52 : ArchMgr has successfully connected to the SpectroSERVER.Jan 26 01:35:52 : ArchMgr has successfully advertised CORBA Event Service.
Jan 26 01:35:52 : ArchMgr is now ready on port 0xbafe, precedence 10
so it is indeed not running at the time of the alarm.
In my case I have the DDM backup on indeed.
thanks for this hint! It looks like this is the reason for this behavior.
Checked it in my test environment and indeed someone enabled the DDM backup within this script.
Once I commented the corresponding lines again, the alarm did not appear any more during an online backup.
Same should be the cause on customer environment, as the DDM backup is enabled as well.