Unsolved
This post is more than 5 years old
72 Posts
0
685094
SanHQ 3.1 false alerts about volume usage
Just upgraded to SanHQ 3.1.0 and I have a 2 issues:
1) Is it normal it reports reports 3.0.9.something and says it’s a beta?
2) After the update, I have a bunch of active alerts about space usage on my luns which weren’t there before (arrays are a mix of 7.0.9, 7.0.10 and 7.1.2 and 8.0.3).
My luns are NOT thin provisioned on the EQL side and there is no snapshot reserve.
When I connect to the group, I see the volume space is showing 100% (which is expected)
In VMware, I see there is free space. I expected the array to report 100% all the time after I filled up enough of the datastore and I was ok with that as I could see the real data in VMware. The problem is SANHQ now shows all my arrays with a red icon because it’s reporting false alerts about space usage.
Is this a known issue? (I didn’t notice anything in the release notes for SanHQ)
Do I have to install VSM to give the array visibility into VMware to report usage correctly and make SanHQ happy?
Is there any way to make SanHQ 3.1 go back to reporting the way 3.0.2 did?
Thanks
crackedup
72 Posts
0
July 17th, 2015 07:00
So it looks like everything I ran into is a known issue. Quick recap...
Sanhq issues
1) member space alert: the alert is correct but the text that accompanies the alert is wrong. Instead of saying, for example, I've crossed the less than 20% free, it says I have less than 0.2% free. The decimal is in the wrong place.
2) changing the value of a threshold doesn't always show the correct value when coming back to that screen (it shows default values).
Trying to change one value ends up resetting all values because they are displaying defaults instead of current value. Seems like you have one chance to change everything and you may not be able to see currently set value in the future. With the new version being more vocal about alerts, changing values becomes a lot more important to avoid false alerts (at least in my environment)
I was told both of these are known issues. The recommendation is to revert to 3.0.2. That's no fun because I lose out on seeing the VVol stuff. Knowing the slow pace of releases, maybe Donald can find out what kind of time frame we're looking at for a real fix? :) Maybe is more people complain, they will push out a fix sooner? :)
VVol issue (also known issue):
I hinted at this in my last comment and this probably isn't the thread for it but I'll put it out there in case it helps someone else.
I storage vmotioned a VM from a VVol to a normal datastore and group manager still shows the vm as being in the storage container with a zero size. Sanhq also shows the vm as still being there (I can partially understand this for historical data perhaps but putting the timeline to a period where the VM isn't there still shows it as being selectable in the list of volumes. )
The workaround for the group manager issue is remove VM from inventory (in vcenter) and re-add it. Not sure if that fixes sanhq issue as well but I don't like the workaround anyway. A fix will come eventually in the form on an updated firmware.
And finally a quick question for Don: can I uninstall sanhq 3.1 and reinstall 3.0.2 or did 3.1 change some of the data/config files that might cause a problem for 3.0.2?
JoshCurrier
4 Posts
0
July 17th, 2015 13:00
I also got confirmation from support that the next version should be out soon, and that it should resolve these issues.
ds281
10 Posts
0
August 13th, 2015 13:00
Confirm this is happening to us with SAN HQ 3.1 (Production release). Volume alerts for critical and warning in-use space, which are false.
Chris.G
40 Posts
0
August 26th, 2015 11:00
Just installed 3.1, and also getting a ton of 2.2 critical notifications, despite restarting the EQLXPerf service and setting thresholds properly. Also a ton of 4.2s on thick provisioned volumes.
Did anyone ever have a fix for this? Seems unusable at this point.
crackedup
72 Posts
0
August 26th, 2015 12:00
The decimal in the wrong place problem has been fixed I believe so the 0.2% free warning makes more sense (ie. 20%).
For the thick provisioned stuff, I just cranked my alerts up to 100% and it doesn't bug me anymore. Problem is I think the editing bug is still there so you have to change all your settings right away otherwise everything will revert when you try to change something again.
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
0
August 26th, 2015 12:00
Hello,
You need to change the values under all the level tabs. Not just critical. The 2.2/4.2 alerts are triggered at different levels.
Regards,
Don
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
0
August 26th, 2015 15:00
On the volumes that are still alarming, how much free space do those volumes show? If you are not actually saving significant space. I.e. the volume is 98% in-use, I would suggest you convert it to a thick provisioned volume. This can be done at any time w/o interrupting ongoing I/O.
Don
Chris.G
40 Posts
0
August 26th, 2015 15:00
Thanks. The 4.2 spam continues despite being set at each level, and threshold being set to 100% at each level. Looks like 2.2 was fixed by setting at multiple levels at least. Not sure about the editing bug. All of the set threshold limits for Voume In-Use Space Percentages still show as 100%, but still alarm at 100%.
Schn1
5 Posts
0
September 1st, 2015 07:00
Hi going through this I noticed that the Free Size calculating does not use the actual Volume size. Instead it uses the DynamicThinReserveMB (value name from the Powershell Tools) value. This Value is calculated out of the Percentage you used while making the thin provisioned Volume (the first one of the 3 percent-Values)
Like here the 57%:
So in this example the SAN-HQ will calculate the FreePercentage like this:
(I took the values showed in the Powershell (Get-EqlVolume)
Volume Size Thin Provisioned = Actual Configured Volume Size * Percent of Size Set
Volume Size Thin Provisioned = 1048576MB * 57% = 597688 MB
Precent Free Calculation of SAN-HQ = Used MB * 100 / Volume Size Thin Provisioned
Precent Free Calculation of SAN-HQ = 542490 * 100 / 597688 = 90,76 %
This is exactly the Value that is shown in my warning mails.
If it was calculating with the actual set size (1TB) thee free percentage would be 51,73.
I wrote a Powershell Script that would calculate a new percentage for all Volumes on the Group that the FreePercentage for thin provisioned volumes will be under a certain value (default 87%)
Also if you like you can uncomment the set-eqlvolume part and it will set the new calculated value. Default I left it read only
Tested with following SW/FW Versions (in 2 Environments)
SAN-HQ 3.1.0
HIT-KIT PSTools 4.8.0
EQL FW 8.0.3
Script can be found here.
Regards,
Florian
harryjohnston
25 Posts
0
September 8th, 2015 21:00
Note that to turn off volume in-use alerts altogether (which is probably what you want to do unless your client operating systems all support trimming) you do not set the threshold to 100%, you remove it - just leave the setting blank. (Or you can disable the entire "volume performance and capacity" alert, ID 4.2, but you might want to use one of the other thresholds but not the volume in-use threshold.)
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
1
September 10th, 2015 11:00
You can also change the interval on the alerts from 6 hours up to 7 days .
Under Settings->Server Settings->Server E-mail Notification Frequency
There's a drop down menu there.
Regards,
Don
srob123
1 Message
0
October 25th, 2015 18:00
I just installed 3.1.0.7890 along with 8.0.5 in my group, and have been noticing this behavior. Even fairly new volumes with relatively little information on them are showing as 95%+ full.
I logged a case with Dell and will update once I know more. I think it's a legitimate bug.
LasseHT
1 Message
0
October 26th, 2015 09:00
Relating to this thread, I installed 3.1 and are now getting alerted in SANHQ regarding our disk usage. Correctly the SAN is tell me that we have allocated almost 100% of the space in fixed volumes, but how can I get rid of this alert as this is as we intended. I rather be able to see (get a red X) if a disk goes bad.
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
0
October 26th, 2015 10:00
Hello,
These issues are going to be addressed in a future release of SANHQ.
Regards,
Don
pjr2010
3 Posts
0
December 18th, 2015 08:00
I opened a case for this issue with EQL support and got the following response
--------
Engineering is aware of this issue in the release of v3.1.0 in SANHQ in relation to thin provisioned volumes.
Actually, its calculating in use space on the basis of Volume reserve, not taking into account the unreserved space.
%in use space= in use space/( in use space + free space)
Unfortunately, the only solution at the moment is to modify (or disable) the alerts for performance and in-use space. In the e-mail settings->Notifications. You’ll probably have to disable it under both “Caution”, “Warning” and “Critical” we are working on this to resolve this issue in next SANHQ release.
-------