2.0.0p17 on Ubuntu 20.04
Hi folks
In October there was an omd update from 1.6.0p5. Everything was running perfectly and smoothly.
Since Dec-05 some strange spikes has been visible:
The TS occupies about 2,3TB, the drive is as small as 3TB, but the spikes even reach more than 4,3TB
Thanks in advance.
Bests,
Darius
Hi @DariusBRB and welcome to the checkmk community.
Do you see these spikes on all oracle TBS or just a few and are they on the day based graphs or also on the less detailed?
You may could provide the relevant output from the agent of the oracle_tablespaces section?
Are you using ASM or normal os storage? Do you cache the TBS checks or are they determined on each agent call?
Hi Tosch,
The database hasn’t change at all,
non ASM, one single line of output looks like these
DB1|C:\DB1\DATAFILE\K2_MF_USERS_GT2MOM84_.DBF|USERS|AVAILABLE|YES|3932160|3932160|3932032|64000|ONLINE|8192|ONLINE|896|PERMANENT|12.1.0.2.0
But maybe the reason could be a new rule: This is a huge TS and I set the warn threshold to 4% free and critical 2%. This rule filters to entries at Explicit hosts and a Explicit tablespaces.
Only one TS is affected, but others on same instance have weird limits (but not any spikes) :
The weird TBS could be UNDO or TEMP TBS which are most of the time empty and if you have set the limits to absolut this may have no sence at the graph level.
I see you have outoextend on at least one data file. I suggest there are index rebuilds ongoing and for the rebuild process indexes need the space twice until the process is finished. this could lead in such spikes. A bit curious i am about the TBS size, which shouldn’t shrink after such a process.
I am quite confident that this issue is not related to check**mk* but rather some suspicious processes at the DB itself.
An i can comfort you, 3TB isn’t that large 
I would also recomment you to set your threasholds accordingly to the TBS sizes in maybe 500GB increments. This is how we deal with different sizes, because 5% free on a 20TB TBS is quite a big bunch of free blocks.
I also have some function in mind like dynamic data management or dynamic space management, i can’t remnember cleary, need to take a look around the web first.
Hi Tosch,
the TBS is just a standard TBS like USERS - not UNDO or TEMP
Yes, while index rebuild more space might be in use - but not at all more than the size of the drive: 3 TB - that’s imho complete impossible.
and there isn’t any magical shrinking. All these datafiles have a fix size (which is the maximum possibly size) - the autoextend ON is an heritage.
Is it possible to get the agent section oracle_tablespace from the time point if these spikes in the graphs happen?
I meant the other instance you posted the picture about. But after thinking through this might be the case because you have set an upper threshold which isn’t total size related.
This topic was automatically closed 365 days after the last reply. New replies are no longer allowed. Contact an admin if you think this should be re-opened.
