Printable View
Hello !
We recently experienced the following in our DSSV6/ESX environment:
Primary/active had a raid controller failure,
the controller didn't die completely,
but couldn't write to his volumes anymore.
So logically the iscsi connections were still up,
but all our VMs on our ESX hosts were frozen.
If a volume becomes physically unwritable on primary/active,
we suggested the system to failover to secondary.
Are we wrong ?
Regards
Ralph
This is correct with the new IO Error detection that we have in DSS V6 from release 5626. But unless there was something else that prevented the failover action then we would need to see the log file from both systems to determine why it did not fail over.
What version of DSS V6 are you using, and what are your advanced settings in the failover configuration?
Thanks for fast reply !
We are DSSV6 up55 (5087).
So I suppose we should upgrade soon.
Since I'm not allowed to post attachments here, should we open a ticket to send logs ?
But if this is fixed from 5626 we just upgrade to 5845 and we'll be fine next time ?
regards
Ralph
update to the latest version, and enable failover on IO errors, and your all set.
Thanks again !
Is it save to upgrade from 5087 directly to 5845 ?
Regards,
Ralph
Yes, its safe.Quote:
Is it save to upgrade from 5087 directly to 5845 ?
Simply download the latest ISO in your user portal, and upload to the server, and reboot.
Thank You !
Got that update already, we have multiple systems running here,
but they were all updated some time before to up80,
before we went to up90.
I'll give it a try.
Looking to our systems already running on up90,
the failover on I/O errors is already enabled by default
so we don't have to configure this.
Good idea, because we could not configure this anyway,
since for configuration changes we would have to stop failover...
Regards,
Ralph
The update from build up55.5087 to up90.5845 should be done with no issues.
Just in case if you face any issue with updateing old system to a newer one, you may ask for a small update ( upd_0756 ) from the support team.
What raid controller are you using and what firmware version is it at?
I've had lots of issues with previous firmware versions (3.66 and 2.00) for HP Smart Array P410 when using SATA disks.
I'm in the same boat and currently running 5087 and want to get to the latest. When I went thru the update check it only found 5626 instead of 5845 as I was expecting. Is there a reason the Update Check isn't going to 5845?
Also will any of these updates affect failover (virtual IP). I notice it said something in the change notes for 5626 that now all IP would have the auxillary connection set for each of you nics, didn't know if that would affect anything.
Thanks!
Andy
Andy,
The auto-updater gets the next build in line.
In your case, just download the latest ISO in your user portal, and upload it manually to the webGUI.
The aux requirement is not in the latest build, so you are ok to move to the latest build.
Just one more question regarding the new feature "failover on I/O error":
What defines the condition "I/O error" ?
Is this condition taken from hardware or from filesystem ?
We wouldn't want a failover when one of the disks in our raid 5 with hot spare reports a repaired sector or similar....
Regards,
Ralph
@Diesoftware:
We've tested these feature and removed (e.g) disk for disk in a raid5 system.
"Degraded" mode (removing one disk) doesn't trigger the failover. Only a failed/broken raid volume leads to it.
This was our experience in our testing environment.
@r3vo
Thanks so far.
What hardware were You using for testing ?
Regards,
Ralph
We made the tests with Adaptec H/W raid controller and a "lsi chip based" controller.
The failover triggered with both controllers when the h/w raid volume got status failed.